Datasets:

ggfox00000
/

dia-ICSIMeetingCorpus-all

	<?xml version="1.0" encoding="iso-8859-1"?>
	<!DOCTYPE Meeting SYSTEM "http://www.icsi.berkeley.edu/Speech/mr/Meeting1.dtd">

	<Meeting Session="Bro003" DateTimeStamp="2000-11-02-1050" Version="2" VersionDate="Jan 7 2004">

	<Preamble>
	<Notes>
	me011 is heard fleetingly during this meeting.
	(He is unmiked, not true meeting participant.)
	</Notes>
	<Channels>
	<Channel Name="chan0" Mic="l1" Transmission="w1" Gain="8" AudioFile="chan0.sph"/>
	<Channel Name="chan1" Mic="s1" Transmission="w1" Gain="8" AudioFile="chan1.sph"/>
	<Channel Name="chan2" Mic="s1" Transmission="w1" Gain="8" AudioFile="chan2.sph"/>
	<Channel Name="chan3" Mic="s1" Transmission="w1" Gain="8" AudioFile="chan3.sph"/>
	<Channel Name="chan4" Mic="s1" Transmission="w1" Gain="8" AudioFile="chan4.sph"/>
	<Channel Name="chan6" Mic="c2" Transmission="u1" Gain="8" AudioFile="chan6.sph"/>
	<Channel Name="chan7" Mic="c2" Transmission="u1" Gain="8" AudioFile="chan7.sph"/>
	<Channel Name="chan8" Mic="p1" Transmission="j1" Gain="11" AudioFile="chan8.sph"/>
	<Channel Name="chan9" Mic="a1" Transmission="j1" Gain="11" AudioFile="chan9.sph"/>
	<Channel Name="chanC" Mic="u3" Transmission="j1" Gain="12" AudioFile="chanC.sph"/>
	<Channel Name="chanD" Mic="u3" Transmission="j1" Gain="12" AudioFile="chanD.sph"/>
	<Channel Name="chanE" Mic="c2" Transmission="u1" Gain="8" AudioFile="chanE.sph"/>
	<Channel Name="chanF" Mic="c2" Transmission="u1" Gain="8" AudioFile="chanF.sph"/>
	</Channels>
	<Participants DB="icsi1.spk">
	<Participant Name="me018" Channel="chan0"/>
	<Participant Name="me034" Channel="chan1"/>
	<Participant Name="me006" Channel="chan2"/>
	<Participant Name="me026" Channel="chan3"/>
	<Participant Name="fn002" Channel="chan4"/>
	<Participant Name="me013" Channel="chan8"/>
	<Participant Name="mn007" Channel="chan9"/>
	<Participant Name="me011"/>
	</Participants>
	</Preamble>


	<Transcript StartTime="0.0" EndTime="5222.894">

	<Segment StartTime="0.000" EndTime="0.876" Participant="me026">
	Channel one.
	</Segment>
	<Segment StartTime="0.278" EndTime="1.225" Participant="mn007">
	Test.
	</Segment>
	<Segment StartTime="0.931" EndTime="1.574" Participant="fn002">
	Hello.
	</Segment>
	<Segment StartTime="0.931" EndTime="3.611" Participant="me034">
	<VocalSound Description="whistling"/>
	</Segment>
	<Segment StartTime="1.488" EndTime="2.122" Participant="me026">
	Channel three.
	</Segment>
	<Segment StartTime="1.586" EndTime="2.456" Participant="mn007">
	Test.
	</Segment>
	<Segment StartTime="1.856" EndTime="2.576" Participant="me018">
	Uh-oh.
	</Segment>
	<Segment StartTime="3.426" EndTime="4.987" Participant="mn007">
	<NonVocalSound Description="mike clicks"/>
	</Segment>
	<Segment StartTime="3.611" EndTime="5.851" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3.635" EndTime="5.635" Participant="me018">
	<NonVocalSound Description="microphone noise"/>
	</Segment>
	<Segment StartTime="5.130" EndTime="6.718" Participant="me013">
	So you think we're going now, yes?
	</Segment>
	<Segment StartTime="5.566" EndTime="6.718" Participant="me034">
	<VocalSound Description="whistling"/>
	</Segment>
	<Segment StartTime="6.718" EndTime="7.380" Participant="me013">
	O_K, good.
	</Segment>
	<Segment StartTime="7.901" EndTime="8.772" Participant="me013">
	Alright
	</Segment>
	<Segment StartTime="8.772" EndTime="9.542" Participant="me013">
	Going again
	</Segment>
	<Segment StartTime="10.179" EndTime="11.200" Participant="me013">
	Uh -
	</Segment>
	<Segment StartTime="11.200" EndTime="14.813" Participant="me013">
	So we're gonna go around as before, and uh do - do our digits.
	</Segment>
	<Segment StartTime="14.169" EndTime="15.415" CloseMic="false">
	<NonVocalSound Description="door closes"/>
	</Segment>
	<Segment StartTime="14.813" EndTime="14.988" Participant="me013">
	<NonVocalSound Description="snap"/>
	</Segment>
	<Segment StartTime="15.632" EndTime="16.119" Participant="me013">
	Uh
	</Segment>
	<Segment StartTime="16.119" EndTime="20.837" Participant="me013" DigitTask="true">
	transcript one three one one dash one three three zero. <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="20.837" EndTime="22.061" Participant="me013" DigitTask="true">
	three two three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="22.679" EndTime="24.413" Participant="me013" DigitTask="true">
	four seven six five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="24.913" EndTime="27.426" Participant="me013" DigitTask="true">
	five three one six two four one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="27.916" EndTime="28.772" Participant="me013" DigitTask="true">
	six seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="29.342" EndTime="29.988" Participant="me013" DigitTask="true">
	seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="30.569" EndTime="30.925" Participant="me013" DigitTask="true">
	eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="31.787" EndTime="34.287" Participant="me013" DigitTask="true">
	nine zero nine four zero zero three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="34.876" EndTime="36.378" Participant="me013" DigitTask="true">
	zero one five eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="37.072" EndTime="38.787" Participant="me013" DigitTask="true">
	one seven three five three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="39.458" EndTime="40.787" Participant="me013" DigitTask="true">
	two six eight zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="41.311" EndTime="43.773" Participant="me013" DigitTask="true">
	three six two four three zero seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="44.216" EndTime="44.931" Participant="me013" DigitTask="true">
	four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="45.439" EndTime="47.509" Participant="me013" DigitTask="true">
	five zero six nine four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="47.813" EndTime="49.035" Participant="me013" DigitTask="true">
	seven four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="49.648" EndTime="50.784" Participant="me013" DigitTask="true">
	eight five seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="51.292" EndTime="52.792" Participant="me013" DigitTask="true">
	nine six one five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="53.212" EndTime="54.908" Participant="me013" DigitTask="true">
	O_ seven eight O_ two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="55.427" EndTime="58.149" Participant="me013" DigitTask="true">
	zero nine six zero four zero zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="58.650" EndTime="59.239" Participant="me013" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="59.752" EndTime="60.219" Participant="me013" DigitTask="true">
	two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="61.929" EndTime="67.889" Participant="me026" DigitTask="true">
	I am reading transcript one three nine one one four one zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="67.889" EndTime="68.566" Participant="me026" DigitTask="true">
	six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="69.135" EndTime="69.723" Participant="me026" DigitTask="true">
	seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="70.361" EndTime="70.655" Participant="me026" DigitTask="true">
	eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="71.730" EndTime="73.545" Participant="me026" DigitTask="true">
	nine O_ six nine eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="74.410" EndTime="76.622" Participant="me026" DigitTask="true">
	zero one three one nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="77.661" EndTime="78.574" Participant="me026" DigitTask="true">
	one six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="79.905" EndTime="81.382" Participant="me026" DigitTask="true">
	two three seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="82.209" EndTime="82.956" Participant="me026" DigitTask="true">
	three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="83.031" EndTime="83.703" Participant="me026" DigitTask="true">
	four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="84.239" EndTime="84.890" Participant="me026" DigitTask="true">
	five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="85.085" EndTime="85.789" Participant="me026" DigitTask="true">
	six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="87.138" EndTime="88.823" Participant="me026" DigitTask="true">
	eight four seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="89.638" EndTime="92.033" Participant="me026" DigitTask="true">
	nine two zero seven five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="93.129" EndTime="95.498" Participant="me026" DigitTask="true">
	O_ three six nine six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="96.720" EndTime="99.910" Participant="me026" DigitTask="true">
	zero nine three one zero zero three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="100.741" EndTime="101.501" Participant="me026" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="102.053" EndTime="102.916" Participant="me026" DigitTask="true">
	two O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="103.707" EndTime="105.424" Participant="me026" DigitTask="true">
	three zero nine seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="106.692" EndTime="109.555" Participant="me026" DigitTask="true">
	five two six seven nine eight three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="110.361" EndTime="112.148" Participant="me026" DigitTask="true">
	six seven zero six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="114.556" EndTime="119.943" Participant="me018" DigitTask="true">
	O_K I've got transcript one two seven one one two nine zero. <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="119.943" EndTime="121.323" Participant="me018" DigitTask="true">
	one seven one O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="122.348" EndTime="124.948" Participant="me018" DigitTask="true">
	two eight one two zero seven one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="125.656" EndTime="126.129" Participant="me018" DigitTask="true">
	three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="126.986" EndTime="127.481" Participant="me018" DigitTask="true">
	four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="128.422" EndTime="131.112" Participant="me018" DigitTask="true">
	five zero nine six zero eight zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="132.338" EndTime="133.241" Participant="me018" DigitTask="true">
	seven three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="134.591" EndTime="136.254" Participant="me018" DigitTask="true">
	eight six eight one three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="137.367" EndTime="139.961" Participant="me018" DigitTask="true">
	nine three four zero three nine four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="140.903" EndTime="142.110" Participant="me018" DigitTask="true">
	O_ four two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="143.051" EndTime="143.498" Participant="me018" DigitTask="true">
	zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="144.201" EndTime="144.656" Participant="me018" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="145.789" EndTime="148.089" Participant="me018" DigitTask="true">
	two zero three eight two three two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="149.648" EndTime="151.485" Participant="me018" DigitTask="true">
	four two eight one six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="152.528" EndTime="154.781" Participant="me018" DigitTask="true">
	five three five eight three O_ eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="156.178" EndTime="159.068" Participant="me018" DigitTask="true">
	six five nine one zero seven six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="160.247" EndTime="161.711" Participant="me018" DigitTask="true">
	seven six six two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="162.880" EndTime="163.708" Participant="me018" DigitTask="true">
	eight seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="164.429" EndTime="164.952" Participant="me018" DigitTask="true">
	nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="166.625" EndTime="169.011" Participant="me018" DigitTask="true">
	O_ O_ five seven eight O_ four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="169.848" EndTime="170.882" Participant="me018" DigitTask="true">
	one four one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="172.824" EndTime="178.165" Participant="me034" DigitTask="true">
	Transcript number one two nine one dash one three one zero. <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="178.165" EndTime="179.800" Participant="me034" DigitTask="true">
	two three nine zero two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="180.294" EndTime="182.018" Participant="me034" DigitTask="true">
	three eight one five two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="182.357" EndTime="184.592" Participant="me034" DigitTask="true">
	four seven four six seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="184.848" EndTime="185.603" Participant="me034" DigitTask="true">
	five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="185.903" EndTime="186.625" Participant="me034" DigitTask="true">
	six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="187.300" EndTime="188.210" Participant="me034" DigitTask="true">
	seven zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="188.574" EndTime="189.325" Participant="me034" DigitTask="true">
	nine four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="190.019" EndTime="191.150" Participant="me034" DigitTask="true">
	O_ one O_ O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="192.300" EndTime="194.173" Participant="me034" DigitTask="true">
	zero seven two six zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="195.732" EndTime="198.640" Participant="me034" DigitTask="true">
	one six zero four six four five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="199.163" EndTime="199.695" Participant="me034" DigitTask="true">
	two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="199.810" EndTime="200.437" Participant="me034" DigitTask="true">
	three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="200.912" EndTime="202.528" Participant="me034" DigitTask="true">
	four zero seven two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="203.251" EndTime="204.659" Participant="me034" DigitTask="true">
	six four four zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="205.104" EndTime="206.817" Participant="me034" DigitTask="true">
	seven four one eight eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="207.500" EndTime="209.562" Participant="me034" DigitTask="true">
	eight eight five six three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="209.781" EndTime="210.608" Participant="me034" DigitTask="true">
	nine eight nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="211.245" EndTime="211.734" Participant="me034" DigitTask="true">
	O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="212.100" EndTime="213.518" Participant="me034" DigitTask="true">
	zero zero three nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="213.612" EndTime="214.002" Participant="me034" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="214.952" EndTime="221.521" Participant="fn002" DigitTask="true">
	Transcript number one one seven one dash one one nine zero. <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="221.521" EndTime="223.828" Participant="fn002" DigitTask="true">
	eight three four three five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="224.211" EndTime="225.921" Participant="fn002" DigitTask="true">
	nine four three nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="226.273" EndTime="227.093" Participant="fn002" DigitTask="true">
	O_ three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="227.386" EndTime="229.941" Participant="fn002" DigitTask="true">
	zero nine six nine one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="230.446" EndTime="230.865" Participant="fn002" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="231.302" EndTime="232.737" Participant="fn002" DigitTask="true">
	two O_ O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="233.041" EndTime="233.678" Participant="fn002" DigitTask="true">
	three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="234.078" EndTime="236.796" Participant="fn002" DigitTask="true">
	five five six O_ eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="237.186" EndTime="241.178" Participant="fn002" DigitTask="true">
	six three zero four zero one one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="241.777" EndTime="244.134" Participant="fn002" DigitTask="true">
	seven eight eight two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="244.743" EndTime="245.355" Participant="fn002" DigitTask="true">
	eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="245.656" EndTime="247.789" Participant="fn002" DigitTask="true">
	nine eight five five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="248.232" EndTime="248.669" Participant="fn002" DigitTask="true">
	O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="248.954" EndTime="250.203" Participant="fn002" DigitTask="true">
	zero zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="250.751" EndTime="252.130" Participant="fn002" DigitTask="true">
	two one four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="252.490" EndTime="254.899" Participant="fn002" DigitTask="true">
	three six seven O_ two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="255.399" EndTime="256.987" Participant="fn002" DigitTask="true">
	four seven nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="257.214" EndTime="260.732" Participant="fn002" DigitTask="true">
	five nine zero five seven three five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="261.483" EndTime="265.336" Participant="fn002" DigitTask="true">
	six seven six nine four six five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="265.703" EndTime="266.295" Participant="fn002" DigitTask="true">
	seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="267.520" EndTime="270.952" Participant="me006">
	O_K, <VocalSound Description="clears throat"/> this is Barry Chen and I am reading transcript
	</Segment>
	<Segment StartTime="270.952" EndTime="275.503" Participant="me006" DigitTask="true">
	one three five one dash one three seven zero, um <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="275.503" EndTime="276.131" Participant="me006" DigitTask="true">
	four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="276.606" EndTime="279.844" Participant="me006" DigitTask="true">
	five O_ seven eight nine two eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="280.370" EndTime="282.979" Participant="me006" DigitTask="true">
	seven one four eight nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="283.355" EndTime="285.713" Participant="me006" DigitTask="true">
	eight five nine three six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="286.273" EndTime="288.602" Participant="me006" DigitTask="true">
	nine seven one five nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="288.984" EndTime="292.122" Participant="me006" DigitTask="true">
	O_ four seven three O_- one O_ O_ <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="292.621" EndTime="293.317" Participant="me006" DigitTask="true">
	zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="293.754" EndTime="294.409" Participant="me006" DigitTask="true">
	one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="295.057" EndTime="296.727" Participant="me006" DigitTask="true">
	two zero six two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="297.278" EndTime="299.677" Participant="me006" DigitTask="true">
	four three four six seven <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="300.266" EndTime="301.780" Participant="me006" DigitTask="true">
	five four four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="301.996" EndTime="302.698" Participant="me006" DigitTask="true">
	six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="303.070" EndTime="305.234" Participant="me006" DigitTask="true">
	seven eight nine nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="305.513" EndTime="306.099" Participant="me006" DigitTask="true">
	eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="306.521" EndTime="307.339" Participant="me006" DigitTask="true">
	nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="307.975" EndTime="309.514" Participant="me006" DigitTask="true">
	O_ O_ one <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="310.009" EndTime="313.013" Participant="me006" DigitTask="true">
	one two four seven three four nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="313.536" EndTime="314.652" Participant="me006" DigitTask="true">
	two five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="314.924" EndTime="316.837" Participant="me006" DigitTask="true">
	three five zero zero eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="317.436" EndTime="318.397" Participant="me006" DigitTask="true">
	four eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="319.239" EndTime="322.460" Participant="mn007" DigitTask="true">
	Transcript <NonVocalSound Description="clicks"/> one three seven one one three nine O_. <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="322.880" EndTime="323.694" Participant="mn007" DigitTask="true">
	five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="323.878" EndTime="324.657" Participant="mn007" DigitTask="true">
	six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="324.838" EndTime="326.327" Participant="mn007" DigitTask="true">
	seven O_ O_ five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="326.649" EndTime="328.251" Participant="mn007" DigitTask="true">
	nine two six three three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="329.116" EndTime="330.297" Participant="mn007" DigitTask="true">
	O_ five five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="330.694" EndTime="331.834" Participant="mn007" DigitTask="true">
	zero four nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="332.433" EndTime="333.329" Participant="mn007" DigitTask="true">
	one six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="333.669" EndTime="334.103" Participant="mn007" DigitTask="true">
	two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="334.363" EndTime="336.920" Participant="mn007" DigitTask="true">
	three O_ O_ two one nine two <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="337.794" EndTime="338.431" Participant="mn007" DigitTask="true">
	four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="338.878" EndTime="339.805" Participant="mn007" DigitTask="true">
	six six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="340.446" EndTime="342.718" Participant="mn007" DigitTask="true">
	seven two O_ eight five O_ three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="343.374" EndTime="344.446" Participant="mn007" DigitTask="true">
	eight six <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="344.707" EndTime="345.365" Participant="mn007" DigitTask="true">
	nine <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="345.822" EndTime="347.662" Participant="mn007" DigitTask="true">
	O_ eight eight seven four <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="348.013" EndTime="348.578" Participant="mn007" DigitTask="true">
	zero <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="349.049" EndTime="350.152" Participant="mn007" DigitTask="true">
	one zero eight <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="350.760" EndTime="352.300" Participant="mn007" DigitTask="true">
	three two four five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="352.908" EndTime="354.052" Participant="mn007" DigitTask="true">
	four five <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="354.344" EndTime="355.979" Participant="mn007" DigitTask="true">
	five six nine two three <Comment Description="Digits"/>
	</Segment>
	<Segment StartTime="357.980" EndTime="358.398" Participant="me013">
	Uh -
	</Segment>
	<Segment StartTime="358.398" EndTime="360.829" Participant="me013">
	Yeah, you don't actually n- need to say the name.
	</Segment>
	<Segment StartTime="360.829" EndTime="362.097" Participant="me013">
	That'll probably be bleeped out.
	</Segment>
	<Segment StartTime="362.353" EndTime="362.838" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="362.353" EndTime="364.077" Participant="me013">
	So. That's if these are anonymized, but
	</Segment>
	<Segment StartTime="364.077" EndTime="364.881" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="364.077" EndTime="366.663" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="364.881" EndTime="366.663" Participant="me006">
	Oh. <Comment Description="mock conspiratorial character voice"/> O_K.
	</Segment>
	<Segment StartTime="366.663" EndTime="367.017" Participant="me013">
	Yeah -
	</Segment>
	<Segment StartTime="367.162" EndTime="368.485" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="367.482" EndTime="372.933" Participant="me013">
	uh - I mean - not that there's anything defamatory about uh - eight five seven or
	</Segment>
	<Segment StartTime="372.933" EndTime="375.692" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="372.933" EndTime="375.692" Participant="me013">
	<VocalSound Description="laugh"/> or anything, but
	</Segment>
	<Segment StartTime="373.248" EndTime="374.354" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="373.340" EndTime="374.403" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="373.372" EndTime="375.380" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="374.039" EndTime="374.984" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="375.692" EndTime="376.670" Participant="me013">
	Uh, anyway.
	</Segment>
	<Segment StartTime="375.692" EndTime="376.317" Participant="me034">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="376.087" EndTime="376.498" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="377.193" EndTime="382.898" Participant="me013">
	Uh - so here's what I have for - I - I was just jotting down things I think th- w- that we should do today.
	</Segment>
	<Segment StartTime="382.898" EndTime="385.904" Participant="me013">
	Uh - This is what I have for an agenda so far
	</Segment>
	<Segment StartTime="386.158" EndTime="386.645" Participant="me013">
	Um,
	</Segment>
	<Segment StartTime="387.425" EndTime="390.257" Participant="me013">
	We should talk a little bit about the plans for the uh -
	</Segment>
	<Segment StartTime="390.257" EndTime="391.829" Participant="me013">
	the field trip next week.
	</Segment>
	<Segment StartTime="391.829" EndTime="393.860" Participant="me013">
	Uh - a number of us are doing a field trip to
	</Segment>
	<Segment StartTime="394.389" EndTime="395.309" Participant="me013">
	uh
	</Segment>
	<Segment StartTime="396.093" EndTime="397.620" Participant="me013">
	Uh - O_G_I
	</Segment>
	<Segment StartTime="397.940" EndTime="399.103" Participant="me013">
	And uh -
	</Segment>
	<Segment StartTime="399.103" EndTime="399.930" Participant="me013">
	mostly uh-
	</Segment>
	<Segment StartTime="399.930" EndTime="401.560" Participant="me013">
	First though about the logistics for it.
	</Segment>
	<Segment StartTime="402.180" EndTime="404.689" Participant="me013">
	Then maybe later on in the meeting we should talk about what we actually
	</Segment>
	<Segment StartTime="404.689" EndTime="406.070" Participant="me013">
	you know, might accomplish.
	</Segment>
	<Segment StartTime="406.053" EndTime="407.484" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="406.070" EndTime="406.950" Participant="me013">
	Uh -
	</Segment>
	<Segment StartTime="407.484" EndTime="408.054" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="408.750" EndTime="412.163" Participant="me013">
	Uh, in and <Pause/> kind of go around - see what people have been doing -
	</Segment>
	<Segment StartTime="412.163" EndTime="415.346" Participant="me013">
	talk about that, <Pause/> a r- progress report. Um,
	</Segment>
	<Segment StartTime="415.396" EndTime="416.356" Participant="me013">
	Essentially.
	</Segment>
	<Segment StartTime="416.840" EndTime="417.240" Participant="me013">
	Um -
	</Segment>
	<Segment StartTime="417.799" EndTime="418.681" Participant="me013">
	And then uh -
	</Segment>
	<Segment StartTime="418.681" EndTime="421.260" Participant="me013">
	Another topic I had was that uh - uh -
	</Segment>
	<Segment StartTime="421.640" EndTime="422.035" Participant="me013">
	Uh -
	</Segment>
	<Segment StartTime="422.170" EndTime="424.832" Participant="me013">
	Dave here had uh said uh "Give me something to do."
	</Segment>
	<Segment StartTime="425.341" EndTime="428.364" Participant="me013">
	And I - I have - I have uh - failed so far in doing that.
	</Segment>
	<Segment StartTime="428.364" EndTime="430.020" Participant="me013">
	And so maybe we can discuss that a little bit.
	</Segment>
	<Segment StartTime="430.020" EndTime="432.120" Participant="me013">
	If we find some holes in some things that - that -
	</Segment>
	<Segment StartTime="432.120" EndTime="434.458" Participant="me013">
	someone could use some help with, he's - he's volunteering to help.
	</Segment>
	<Segment StartTime="434.981" EndTime="436.524" Participant="me018">
	I've got to move a bunch of furniture. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="436.411" EndTime="438.532" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="438.532" EndTime="441.413" Participant="me013">
	O_K, always count on a <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="441.413" EndTime="442.512" Participant="me013">
	serious comment from that corner.
	</Segment>
	<Segment StartTime="442.554" EndTime="443.690" Participant="me013">
	So, um,
	</Segment>
	<Segment StartTime="444.878" EndTime="447.241" Participant="me013">
	uh, and uh, then uh,
	</Segment>
	<Segment StartTime="447.241" EndTime="449.911" Participant="me013">
	talk a little bit about - about disks and resource -
	</Segment>
	<Segment StartTime="449.911" EndTime="452.231" Participant="me013">
	resource issues that - that's starting to get worked out.
	</Segment>
	<Segment StartTime="452.231" EndTime="453.298" Participant="me013">
	And then,
	</Segment>
	<Segment StartTime="453.298" EndTime="455.440" Participant="me013">
	anything else anybody has that
	</Segment>
	<Segment StartTime="455.440" EndTime="457.011" Participant="me013">
	isn't in that list?
	</Segment>
	<Segment StartTime="457.689" EndTime="458.462" Participant="me013">
	Uh -
	</Segment>
	<Segment StartTime="459.093" EndTime="462.522" Participant="me026">
	I was just wondering, does this mean the battery's dying and I should change it?
	</Segment>
	<Segment StartTime="463.525" EndTime="464.211" Participant="me013">
	Uh
	</Segment>
	<Segment StartTime="465.659" EndTime="467.901" Participant="me013">
	I think that means the battery's O_<Emphasis> K. </Emphasis> - d - do you
	</Segment>
	<Segment StartTime="467.344" EndTime="467.849" Participant="me018">
	Let me see.
	</Segment>
	<Segment StartTime="468.261" EndTime="469.190" Participant="me026">
	Oh O_K, so th-
	</Segment>
	<Segment StartTime="469.554" EndTime="470.281" Participant="me018">
	Yeah, that's good.
	</Segment>
	<Segment StartTime="470.649" EndTime="471.086" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="470.662" EndTime="471.141" Participant="me018">
	You're alright?
	</Segment>
	<Segment StartTime="471.037" EndTime="471.769" Participant="me026">
	Cuz it's full.
	</Segment>
	<Segment StartTime="471.086" EndTime="471.669" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="472.073" EndTime="472.782" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="472.168" EndTime="472.568" Participant="me026">
	Alright.
	</Segment>
	<Segment StartTime="472.782" EndTime="473.466" Participant="me013">
	Yeah. It looks
	</Segment>
	<Segment StartTime="473.466" EndTime="474.758" Participant="me013">
	<Emphasis> full </Emphasis> of electrons.
	</Segment>
	<Segment StartTime="474.618" EndTime="476.230" Participant="me026">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="474.758" EndTime="475.948" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="475.948" EndTime="476.643" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="476.643" EndTime="477.998" Participant="me013">
	Plenty of electrons left there.
	</Segment>
	<Segment StartTime="477.998" EndTime="482.200" Participant="me013">
	O_K, so, um, uh. O_K,
	</Segment>
	<Segment StartTime="482.200" EndTime="483.907" Participant="me013">
	so, uh,
	</Segment>
	<Segment StartTime="483.907" EndTime="486.066" Participant="me013">
	I wanted to start this with this mundane thing. Um -
	</Segment>
	<Segment StartTime="486.066" EndTime="488.617" Participant="me013">
	Uh - I - I - it was - it was
	</Segment>
	<Segment StartTime="488.617" EndTime="493.678" Participant="me013">
	kind of <Emphasis> my </Emphasis> bright idea to have us take a plane that leaves at seven twenty in the morning.
	</Segment>
	<Segment StartTime="493.678" EndTime="498.374" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="493.678" EndTime="494.639" Participant="me013">
	<VocalSound Description="inbreath"/> Um.
	</Segment>
	<Segment StartTime="494.612" EndTime="496.196" Participant="me006">
	Oh, yeah, that's right.
	</Segment>
	<Segment StartTime="495.785" EndTime="498.374" Participant="me013">
	Uh <VocalSound Description="laugh"/> this is uh -
	</Segment>
	<Segment StartTime="498.374" EndTime="505.063" Participant="me013">
	The reason I did it uh was because otherwise for those of us who have to come back the same day it is really not much of a -
	</Segment>
	<Segment StartTime="505.063" EndTime="505.900" Participant="me013">
	of a visit.
	</Segment>
	<Segment StartTime="505.900" EndTime="506.911" Participant="me013">
	Uh - <VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="506.911" EndTime="510.839" Participant="me013">
	So um the issue is how - how - how would we ever accomplish that?
	</Segment>
	<Segment StartTime="510.839" EndTime="514.190" Participant="me013">
	Uh - what - what - what part of town do you live in?
	</Segment>
	<Segment StartTime="510.839" EndTime="514.190" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="514.431" EndTime="520.956" Participant="me006">
	Um, I live in, um, the corner of campus. The, um, southeast corner.
	</Segment>
	<Segment StartTime="518.325" EndTime="519.023" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="520.983" EndTime="521.599" Participant="me013">
	O_K, so
	</Segment>
	<Segment StartTime="521.996" EndTime="523.168" Participant="me013">
	would it be easier -
	</Segment>
	<Segment StartTime="523.769" EndTime="524.808" Participant="me013">
	those of you who are not,
	</Segment>
	<Segment StartTime="524.808" EndTime="528.083" Participant="me013">
	you know, used to this area, it can be very tricky to get to the airport at -
	</Segment>
	<Segment StartTime="528.083" EndTime="529.651" Participant="me013">
	at uh, you know, six thirty.
	</Segment>
	<Segment StartTime="529.651" EndTime="532.103" Participant="me013">
	<VocalSound Description="inbreath"/> Um. So.
	</Segment>
	<Segment StartTime="532.390" EndTime="536.318" Participant="me013">
	Would it be easier for you if you came here and I drove you?
	</Segment>
	<Segment StartTime="538.680" EndTime="539.206" Participant="me013">
	Yeah?
	</Segment>
	<Segment StartTime="539.917" EndTime="540.712" Participant="mn007">
	Yeah, perhaps, yeah.
	</Segment>
	<Segment StartTime="540.238" EndTime="541.723" Participant="me013">
	Yeah, yeah, O_K. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="540.778" EndTime="541.331" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="540.863" EndTime="541.200" Participant="mn007">
	Mm-hmm.
	</Segment>
	<Segment StartTime="540.992" EndTime="541.920" Participant="me006">
	Yeah. Sure.
	</Segment>
	<Segment StartTime="541.723" EndTime="545.220" Participant="me013">
	O_K, so if - if everybody can get here at six.
	</Segment>
	<Segment StartTime="541.743" EndTime="542.879" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="545.519" EndTime="546.410" Participant="fn002">
	At six.
	</Segment>
	<Segment StartTime="546.410" EndTime="547.131" Participant="me013">
	Yeah, I'm afraid
	</Segment>
	<Segment StartTime="547.632" EndTime="549.816" Participant="me013">
	we need to do that to get there on time. Yeah, so.
	</Segment>
	<Segment StartTime="548.010" EndTime="549.816" Participant="me006">
	Six, O_K.
	</Segment>
	<Segment StartTime="549.816" EndTime="550.940" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="551.620" EndTime="554.281" Participant="me013">
	<VocalSound Description="inbreath"/> Oh boy. Anyway, so.
	</Segment>
	<Segment StartTime="554.447" EndTime="555.820" Participant="me018">
	Will that <Pause/> be enough time?
	</Segment>
	<Segment StartTime="556.379" EndTime="556.742" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="556.742" EndTime="559.602" Participant="me013">
	Yeah, so I'll just pull up in front at six and just be out front.
	</Segment>
	<Segment StartTime="561.500" EndTime="566.347" Participant="me013">
	And, uh, and yeah, that'll be plenty of time. It'll take - it - it - it won't be bad traffic that time of day and -
	</Segment>
	<Segment StartTime="566.347" EndTime="567.095" Participant="me013">
	and uh
	</Segment>
	<Segment StartTime="566.582" EndTime="570.655" Participant="me018">
	I guess once you get past the bridge <Pause/> that that would be the worst.
	</Segment>
	<Segment StartTime="569.964" EndTime="570.833" Participant="me013">
	Going to Oakland.
	</Segment>
	<Segment StartTime="570.249" EndTime="571.304" Participant="me034">
	<Uncertain> Yeah </Uncertain>, Oakland.
	</Segment>
	<Segment StartTime="570.868" EndTime="572.160" Participant="me006">
	Oakland.
	</Segment>
	<Segment StartTime="570.919" EndTime="571.364" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="571.858" EndTime="572.428" Participant="me013">
	Bridge
	</Segment>
	<Segment StartTime="571.954" EndTime="574.482" Participant="me018">
	Once you get past the turnoff to the <Pause/> Bay Bridge.
	</Segment>
	<Segment StartTime="573.149" EndTime="574.503" Participant="me013">
	oh, the turnoff to the bridge
	</Segment>
	<Segment StartTime="574.554" EndTime="574.826" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="575.360" EndTime="577.830" Participant="me013">
	Won't even do that. I mean, just go down Martin Luther King.
	</Segment>
	<Segment StartTime="576.113" EndTime="576.456" Participant="me034">
	Yeah.
	</Segment>
	<Segment StartTime="577.472" EndTime="578.760" Participant="me018">
	Yeah. O_K. Mm-hmm.
	</Segment>
	<Segment StartTime="577.830" EndTime="582.252" Participant="me013">
	<Pause/> And then Martin Luther King to nine-eighty to eight-eighty, and it's - it'd take us,
	</Segment>
	<Segment StartTime="580.420" EndTime="580.920" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="582.252" EndTime="584.982" Participant="me013">
	tops uh thirty minutes to get there.
	</Segment>
	<Segment StartTime="583.810" EndTime="584.477" Participant="me018">
	Oh, I -
	</Segment>
	<Segment StartTime="585.180" EndTime="587.050" Participant="me013">
	So that leaves us fifty minutes before the
	</Segment>
	<Segment StartTime="587.939" EndTime="589.271" Participant="me013">
	plane - it'll just - yeah.
	</Segment>
	<Segment StartTime="589.680" EndTime="590.069" Participant="me013">
	So
	</Segment>
	<Segment StartTime="590.516" EndTime="592.021" Participant="me013">
	Great, O_K so that'll
	</Segment>
	<Segment StartTime="592.021" EndTime="595.233" Participant="me013">
	<VocalSound Description="inbreath"/> It's - I mean, it's still not going to be really easy but - well
	</Segment>
	<Segment StartTime="595.233" EndTime="596.785" Participant="me013">
	Particularly for - for uh -
	</Segment>
	<Segment StartTime="596.785" EndTime="598.556" Participant="me013">
	for Barry and me, we're not - we're not staying overnight
	</Segment>
	<Segment StartTime="598.556" EndTime="600.991" Participant="me013">
	so we don't need to bring anything particularly except for
	</Segment>
	<Segment StartTime="600.991" EndTime="602.611" Participant="me013">
	<VocalSound Description="inbreath"/> uh -
	</Segment>
	<Segment StartTime="602.611" EndTime="603.369" Participant="me013">
	a pad of paper <VocalSound Description="laugh"/> and -
	</Segment>
	<Segment StartTime="603.369" EndTime="604.320" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="603.419" EndTime="605.571" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="604.320" EndTime="606.031" Participant="me013">
	So, and, uh you, two
	</Segment>
	<Segment StartTime="606.008" EndTime="607.115" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="606.031" EndTime="607.410" Participant="me013">
	have to bring a little bit
	</Segment>
	<Segment StartTime="607.410" EndTime="607.971" Participant="me013">
	but uh -
	</Segment>
	<Segment StartTime="607.971" EndTime="610.363" Participant="me013">
	you know, don't - don't bring a footlocker and we'll be O_K
	</Segment>
	<Segment StartTime="610.363" EndTime="613.260" Participant="me013">
	<VocalSound Description="inbreath"/> So. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="611.380" EndTime="612.372" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="613.260" EndTime="613.630" Participant="me006">
	s-
	</Segment>
	<Segment StartTime="613.577" EndTime="618.002" Participant="me013">
	W- you're staying overnight. I figured you wouldn't need a great big suitcase, yeah. That's sort of <Pause/> <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="613.630" EndTime="614.117" Participant="me006">
	So just -
	</Segment>
	<Segment StartTime="615.643" EndTime="616.984" Participant="mn007">
	Oh yeah. Yeah.
	</Segment>
	<Segment StartTime="616.984" EndTime="617.728" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="618.002" EndTime="620.251" Participant="me013">
	one night. So. Anyway.
	</Segment>
	<Segment StartTime="620.689" EndTime="622.624" Participant="me006">
	So, s- six A_M, in front.
	</Segment>
	<Segment StartTime="620.689" EndTime="622.412" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="622.412" EndTime="623.420" Participant="me013">
	Six A_M in front.
	</Segment>
	<Segment StartTime="623.552" EndTime="623.909" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="623.980" EndTime="625.190" Participant="me013">
	Uh, I'll be here.
	</Segment>
	<Segment StartTime="625.190" EndTime="627.540" Participant="me013">
	<Pause/> Uh - I'll - I'll - I'll -
	</Segment>
	<Segment StartTime="627.540" EndTime="628.721" Participant="me013">
	I'll give you my phone number,
	</Segment>
	<Segment StartTime="628.721" EndTime="631.992" Participant="me013">
	<VocalSound Description="laugh"/> If I'm not here for a few m- after a few minutes then <Pause/>
	</Segment>
	<Segment StartTime="629.193" EndTime="633.756" Participant="me006">
	<VocalSound Description="laugh"/> Wake you up. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="631.992" EndTime="632.878" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="632.878" EndTime="634.803" Participant="me013">
	Nah, I'll be fine. I just, uh -
	</Segment>
	<Segment StartTime="634.803" EndTime="637.062" Participant="me013">
	it - for me it just means getting up a half an hour earlier
	</Segment>
	<Segment StartTime="637.062" EndTime="637.773" Participant="me013">
	than I usually do.
	</Segment>
	<Segment StartTime="637.773" EndTime="640.142" Participant="me013">
	Not - not - not a lot, so
	</Segment>
	<Segment StartTime="639.328" EndTime="639.785" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="640.331" EndTime="642.091" Participant="me006">
	Wednesday. <Comment Description="whispered, lengthened, as if writing it down"/>
	</Segment>
	<Segment StartTime="641.400" EndTime="642.071" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="642.071" EndTime="645.320" Participant="me013">
	O_K, that was the real real important stuff. Um,
	</Segment>
	<Segment StartTime="645.320" EndTime="647.641" Participant="me013">
	I - I - I figured maybe wait on the
	</Segment>
	<Segment StartTime="647.641" EndTime="649.661" Participant="me013">
	potential goals for the meeting
	</Segment>
	<Segment StartTime="649.661" EndTime="650.550" Participant="me013">
	uh - until
	</Segment>
	<Segment StartTime="650.550" EndTime="652.811" Participant="me013">
	we talk about wh- what's been going on.
	</Segment>
	<Segment StartTime="652.901" EndTime="653.536" Participant="me013">
	So,
	</Segment>
	<Segment StartTime="654.290" EndTime="657.800" Participant="me013">
	uh, what's been going on? Why don't we start - start over here.
	</Segment>
	<Segment StartTime="658.276" EndTime="659.375" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="659.821" EndTime="659.901" Participant="mn007">
	<VocalSound Description="mouth"/>
	</Segment>
	<Segment StartTime="659.901" EndTime="663.736" Participant="mn007">
	Well, preparation of the French test data actually. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="663.736" EndTime="664.475" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="664.475" EndTime="665.716" Participant="mn007">
	So,
	</Segment>
	<Segment StartTime="667.046" EndTime="667.823" Participant="mn007">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="667.823" EndTime="669.040" Participant="mn007">
	it means that um,
	</Segment>
	<Segment StartTime="669.040" EndTime="670.290" Participant="mn007">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="670.290" EndTime="673.900" Participant="mn007">
	well, it is, uh, a digit French database of microphone speech,
	</Segment>
	<Segment StartTime="674.650" EndTime="676.999" Participant="mn007">
	downsampled to eight kilohertz and
	</Segment>
	<Segment StartTime="677.190" EndTime="679.408" Participant="mn007">
	I've added noise to one part,
	</Segment>
	<Segment StartTime="680.026" EndTime="681.930" Participant="mn007">
	with the - actually the
	</Segment>
	<Segment StartTime="681.930" EndTime="683.780" Participant="mn007">
	Aurora-two noises. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="684.781" EndTime="685.692" Participant="mn007">
	And, <Uncertain NumSyllables="1"> @@ </Uncertain>
	</Segment>
	<Segment StartTime="686.050" EndTime="687.740" Participant="mn007">
	so this is a training part.
	</Segment>
	<Segment StartTime="687.740" EndTime="691.681" Participant="mn007">
	And then <Pause/> the remaining part, I use for testing and -
	</Segment>
	<Segment StartTime="691.681" EndTime="694.626" Participant="mn007">
	with other kind of noises. So we can -
	</Segment>
	<Segment StartTime="700.834" EndTime="706.193" Participant="mn007">
	So this is almost ready. I'm preparing the - the H_T_K baseline for this task.
	</Segment>
	<Segment StartTime="706.677" EndTime="707.744" Participant="mn007">
	And, yeah.
	</Segment>
	<Segment StartTime="709.184" EndTime="709.777" Participant="me013">
	O_K
	</Segment>
	<Segment StartTime="711.500" EndTime="713.050" Participant="me013">
	Uh,
	</Segment>
	<Segment StartTime="713.050" EndTime="714.791" Participant="me013">
	So the H_T_K base lines -
	</Segment>
	<Segment StartTime="714.791" EndTime="717.095" Participant="me013">
	so this is using mel cepstra and so on, or - ?
	</Segment>
	<Segment StartTime="717.154" EndTime="717.442" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="717.598" EndTime="718.117" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="719.190" EndTime="719.759" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="720.509" EndTime="723.830" Participant="me013">
	And again, I guess the p- the plan is, uh,
	</Segment>
	<Segment StartTime="724.780" EndTime="725.822" Participant="me013">
	to uh -
	</Segment>
	<Segment StartTime="726.910" EndTime="728.032" Participant="me013">
	then given this -
	</Segment>
	<Segment StartTime="729.040" EndTime="731.050" Participant="me013">
	What's the plan again? <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="731.590" EndTime="733.171" Participant="mn007">
	The plan with <Pause/> these data?
	</Segment>
	<Segment StartTime="732.468" EndTime="735.452" Participant="me013">
	With - So - So - Does i- Just remind me of what - what you were going to do
	</Segment>
	<Segment StartTime="735.452" EndTime="736.920" Participant="me013">
	with the - what - what - what - what's - y-
	</Segment>
	<Segment StartTime="736.920" EndTime="738.651" Participant="me013">
	You just described what you've <Emphasis> been </Emphasis> doing.
	</Segment>
	<Segment StartTime="738.441" EndTime="738.737" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="738.651" EndTime="740.847" Participant="me013">
	So if you could remind me of what you're <Emphasis> going </Emphasis> to be doing.
	</Segment>
	<Segment StartTime="740.847" EndTime="742.372" Participant="me013">
	Oh, this is - yeah, yeah.
	</Segment>
	<Segment StartTime="740.990" EndTime="742.301" Participant="mn007">
	Uh, yeah.
	</Segment>
	<Segment StartTime="741.747" EndTime="743.818" Participant="me006">
	<Uncertain NumSyllables="2"> @@ </Uncertain> Tell him about the cube.
	</Segment>
	<Segment StartTime="742.301" EndTime="743.540" Participant="mn007">
	Well.
	</Segment>
	<Segment StartTime="744.234" EndTime="744.861" Participant="mn007">
	The cube?
	</Segment>
	<Segment StartTime="744.861" EndTime="747.392" Participant="mn007">
	I should tell him about the cube? <Comment Description="while laughing"/> Yeah.
	</Segment>
	<Segment StartTime="745.550" EndTime="748.052" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="746.481" EndTime="747.140" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="746.772" EndTime="747.691" Participant="me013">
	Oh! <Emphasis> Cube. </Emphasis> <Emphasis> Yeah. </Emphasis>
	</Segment>
	<Segment StartTime="747.986" EndTime="749.851" Participant="mn007">
	Uh we - actually we want to, mmm,
	</Segment>
	<Segment StartTime="748.052" EndTime="749.461" Participant="fn002">
	Fill in the cube. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="750.064" EndTime="753.050" Participant="mn007">
	Uh, <VocalSound Description="laugh"/> uh,
	</Segment>
	<Segment StartTime="751.649" EndTime="753.050" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="753.050" EndTime="754.659" Participant="mn007">
	analyze three dimensions, <Comment Description="colon"/>
	</Segment>
	<Segment StartTime="755.288" EndTime="759.871" Participant="mn007">
	the feature dimension, the <Pause/> training data dimension, and the test data dimension.
	</Segment>
	<Segment StartTime="761.592" EndTime="762.581" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="764.159" EndTime="766.180" Participant="mn007">
	<VocalSound Description="mouth"/> Well, what we want to do is
	</Segment>
	<Segment StartTime="766.632" EndTime="770.002" Participant="mn007">
	first we have number for each <Pause/> uh task.
	</Segment>
	<Segment StartTime="771.170" EndTime="772.745" Participant="mn007">
	So we have the um,
	</Segment>
	<Segment StartTime="773.180" EndTime="777.807" Participant="mn007">
	T_I-digit task, the Italian task, the French task <Pause/> and the Finnish task. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="777.923" EndTime="778.376" Participant="me013">
	Yeah?
	</Segment>
	<Segment StartTime="777.987" EndTime="780.960" Participant="mn007">
	So we have numbers with <Pause/> uh - systems -
	</Segment>
	<Segment StartTime="780.960" EndTime="781.980" Participant="mn007">
	I mean - I mean
	</Segment>
	<Segment StartTime="781.980" EndTime="784.969" Participant="mn007">
	neural networks trained on the task data. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="786.816" EndTime="790.575" Participant="mn007">
	And then to have systems with neural networks trained on,
	</Segment>
	<Segment StartTime="791.023" EndTime="792.012" Participant="mn007">
	<VocalSound Description="mouth"/> uh,
	</Segment>
	<Segment StartTime="792.012" EndTime="795.264" Participant="mn007">
	data from the same language, if possible, with,
	</Segment>
	<Segment StartTime="795.264" EndTime="799.756" Participant="mn007">
	<Uncertain> well, </Uncertain> using a more generic database, which is phonetically - phonetically balanced, and.
	</Segment>
	<Segment StartTime="801.181" EndTime="802.423" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="805.957" EndTime="806.629" Participant="mn007">
	Yeah. So.
	</Segment>
	<Segment StartTime="805.957" EndTime="809.140" Participant="me013">
	So- so we had talked - I guess we had talked at one point about
	</Segment>
	<Segment StartTime="809.140" EndTime="811.420" Participant="me013">
	maybe, the language I_D corpus?
	</Segment>
	<Segment StartTime="812.308" EndTime="814.240" Participant="me013">
	Is that a possibility for that?
	</Segment>
	<Segment StartTime="812.645" EndTime="813.521" Participant="mn007">
	Ye- uh - <Pause/>
	</Segment>
	<Segment StartTime="813.521" EndTime="818.997" Participant="mn007">
	Yeah, but, uh these corpus, w- w- there is a CallHome and a CallFriend also,
	</Segment>
	<Segment StartTime="818.997" EndTime="822.934" Participant="mn007">
	The CallFriend is for language ind- identification. Well,
	</Segment>
	<Segment StartTime="823.606" EndTime="827.510" Participant="mn007">
	anyway, these corpus are all telephone speech. So, um.
	</Segment>
	<Segment StartTime="828.331" EndTime="829.240" Participant="mn007">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="829.240" EndTime="831.784" Participant="mn007">
	This could be a - <Pause/> a problem for -
	</Segment>
	<Segment StartTime="834.200" EndTime="836.511" Participant="mn007">
	Why? Because uh, uh,
	</Segment>
	<Segment StartTime="836.511" EndTime="840.194" Participant="mn007">
	the - the SpeechDat databases are not telephone speech.
	</Segment>
	<Segment StartTime="840.560" EndTime="842.070" Participant="mn007">
	They are downsampled to eight kilohertz but -
	</Segment>
	<Segment StartTime="842.070" EndTime="842.760" Participant="mn007">
	but
	</Segment>
	<Segment StartTime="842.760" EndTime="845.988" Participant="mn007">
	they are not <VocalSound Description="inbreath"/> uh with telephone bandwidth.
	</Segment>
	<Segment StartTime="844.671" EndTime="845.775" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="845.775" EndTime="846.950" Participant="me013">
	That's really funny isn't it?
	</Segment>
	<Segment StartTime="846.950" EndTime="851.273" Participant="me013">
	I mean cuz th- this whole thing is for <Pause/> developing new standards for the telephone.
	</Segment>
	<Segment StartTime="849.051" EndTime="850.274" Participant="me006">
	Telephone. <Comment Description="completing [me013]'s utterance"/> <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="851.273" EndTime="851.886" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="852.230" EndTime="855.942" Participant="mn007">
	Yeah, but the - the idea is to compute the feature before <Pause/> the -
	</Segment>
	<Segment StartTime="855.942" EndTime="857.768" Participant="mn007">
	before sending them to the -
	</Segment>
	<Segment StartTime="857.768" EndTime="865.500" Participant="mn007">
	Well, <Pause/> you don't - do not send speech, you send features, computed on th- the - <Pause/> the device, or - Well.
	</Segment>
	<Segment StartTime="861.403" EndTime="862.059" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="863.470" EndTime="864.540" Participant="me013">
	Yeah, I know, but the reason -
	</Segment>
	<Segment StartTime="866.701" EndTime="868.531" Participant="me013">
	Oh I see, so your point is that it's - it's -
	</Segment>
	<Segment StartTime="867.037" EndTime="867.929" Participant="mn007">
	So <Pronounce Pronunciation=""yyyou""> you </Pronounce> -
	</Segment>
	<Segment StartTime="868.531" EndTime="869.790" Participant="me013">
	it's uh -
	</Segment>
	<Segment StartTime="869.790" EndTime="874.490" Participant="me013">
	the features are computed locally, and so they aren't necessarily telephone bandwidth, uh
	</Segment>
	<Segment StartTime="872.016" EndTime="872.953" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="873.732" EndTime="874.270" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="875.410" EndTime="876.482" Participant="me013">
	or telephone
	</Segment>
	<Segment StartTime="876.505" EndTime="881.027" Participant="me018">
	Did you <Pause/> happen to find out anything about the O_G_I multilingual database?
	</Segment>
	<Segment StartTime="877.065" EndTime="877.866" Participant="me013">
	distortions.
	</Segment>
	<Segment StartTime="880.782" EndTime="882.628" Participant="me013">
	Yeah, that's wh- that's wh- that's what I <Emphasis> meant. </Emphasis>
	</Segment>
	<Segment StartTime="880.903" EndTime="881.884" Participant="mn007">
	Yeah, it's -
	</Segment>
	<Segment StartTime="882.628" EndTime="886.528" Participant="me013">
	I said - <Uncertain> @@ </Uncertain>, there's - there's - there's an O_G_I language I_D, not the - not the, uh -
	</Segment>
	<Segment StartTime="886.528" EndTime="890.717" Participant="me013">
	the CallFriend is a - is a, uh, L_D_C w- thing, right?
	</Segment>
	<Segment StartTime="887.893" EndTime="888.216" Participant="mn007">
	Yea-
	</Segment>
	<Segment StartTime="888.798" EndTime="891.412" Participant="mn007">
	Yeah, there are also two other databases. <Comment Description="colon"/>
	</Segment>
	<Segment StartTime="891.412" EndTime="894.224" Participant="mn007">
	One they call the multi-language database,
	</Segment>
	<Segment StartTime="894.224" EndTime="897.357" Participant="mn007">
	and another one is a twenty-two language, something like that.
	</Segment>
	<Segment StartTime="897.357" EndTime="899.870" Participant="mn007">
	But it's also telephone speech.
	</Segment>
	<Segment StartTime="899.328" EndTime="899.983" Participant="me018">
	Oh, they are?
	</Segment>
	<Segment StartTime="900.129" EndTime="900.807" Participant="me018">
	O_K.
	</Segment>
	<Segment StartTime="900.807" EndTime="901.691" Participant="mn007">
	Uh.
	</Segment>
	<Segment StartTime="903.330" EndTime="903.740" Participant="mn007">
	Well,
	</Segment>
	<Segment StartTime="904.436" EndTime="905.732" Participant="mn007">
	nnn.
	</Segment>
	<Segment StartTime="905.766" EndTime="907.231" Participant="me013">
	But I'm not sure -
	</Segment>
	<Segment StartTime="906.160" EndTime="906.778" Participant="mn007">
	So -
	</Segment>
	<Segment StartTime="908.284" EndTime="912.002" Participant="me013">
	I mean, we'r- e- e- The bandwidth shouldn't be such an issue right? Because
	</Segment>
	<Segment StartTime="912.002" EndTime="914.712" Participant="me013">
	e- e- this is downsampled and - and filtered, right?
	</Segment>
	<Segment StartTime="914.712" EndTime="917.010" Participant="me013">
	So it's just the fact that it's not telephone.
	</Segment>
	<Segment StartTime="917.399" EndTime="928.573" Participant="me013">
	And there are so many other differences between these different databases. I mean some of this stuff's recorded in the car, and some of it's - I mean there's - there's many different acoustic differences. So I'm not sure
	</Segment>
	<Segment StartTime="926.982" EndTime="927.481" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="929.570" EndTime="930.570" Participant="me013">
	if - .
	</Segment>
	<Segment StartTime="930.570" EndTime="934.978" Participant="me013">
	I mean, unless we're going to include a bunch of car recordings in the - in the training database,
	</Segment>
	<Segment StartTime="934.978" EndTime="937.253" Participant="me013">
	I'm not sure if it's -
	</Segment>
	<Segment StartTime="937.253" EndTime="938.580" Participant="me013">
	completely rules it out
	</Segment>
	<Segment StartTime="938.814" EndTime="943.295" Participant="me013">
	if our - if we - if our major goal is to have phonetic context
	</Segment>
	<Segment StartTime="943.541" EndTime="944.068" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="943.810" EndTime="946.988" Participant="me013">
	and you figure that there's gonna be a mismatch in acoustic conditions
	</Segment>
	<Segment StartTime="947.420" EndTime="949.750" Participant="me013">
	does it make it much worse f- to
	</Segment>
	<Segment StartTime="949.750" EndTime="951.571" Participant="me013">
	sort of add another mismatch, if you will.
	</Segment>
	<Segment StartTime="952.106" EndTime="958.001" Participant="me013">
	Uh, i- i- I - I guess the question is how important is it to - for us to get multiple languages
	</Segment>
	<Segment StartTime="958.700" EndTime="960.008" Participant="me013">
	uh, in there.
	</Segment>
	<Segment StartTime="959.771" EndTime="960.345" Participant="mn007">
	Yeah, but -
	</Segment>
	<Segment StartTime="960.986" EndTime="962.122" Participant="mn007">
	Mm-hmm. <VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="962.350" EndTime="963.304" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="964.629" EndTime="965.143" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="966.382" EndTime="975.368" Participant="mn007">
	Well, actually, for the moment if we w- do not want to use these phone databases, we - we already have uh - English, Spanish and French
	</Segment>
	<Segment StartTime="973.071" EndTime="974.045" Participant="me034">
	<VocalSound Description="cough"/>
	</Segment>
	<Segment StartTime="975.368" EndTime="977.283" Participant="mn007">
	uh, with microphone speech.
	</Segment>
	<Segment StartTime="975.420" EndTime="975.969" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="977.142" EndTime="977.656" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="977.283" EndTime="977.716" Participant="mn007">
	So.
	</Segment>
	<Segment StartTime="979.580" EndTime="982.997" Participant="me013">
	So that's what you're thinking of using is sort of the multi- the equivalent of the multiple?
	</Segment>
	<Segment StartTime="979.918" EndTime="980.366" Participant="mn007">
	Well.
	</Segment>
	<Segment StartTime="981.999" EndTime="986.274" Participant="mn007">
	Yeah, for the multilingual part we were thinking of using these three databases.
	</Segment>
	<Segment StartTime="986.650" EndTime="989.916" Participant="me013">
	And for the difference in phonetic context <Pause/> that you - ?
	</Segment>
	<Segment StartTime="986.720" EndTime="987.383" Participant="mn007">
	Well, this -
	</Segment>
	<Segment StartTime="989.916" EndTime="993.681" Participant="mn007">
	Uh, actually, these three databases are um generic databases.
	</Segment>
	<Segment StartTime="992.282" EndTime="993.041" Participant="me013">
	Provide that. <Comment Description="completing [mn007]'s utterance"/>
	</Segment>
	<Segment StartTime="993.010" EndTime="993.393" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="993.942" EndTime="995.098" Participant="mn007">
	So w-
	</Segment>
	<Segment StartTime="995.098" EndTime="997.662" Participant="mn007">
	f- for - for uh Italian,
	</Segment>
	<Segment StartTime="998.077" EndTime="1002.410" Participant="mn007">
	which is close to Spanish, French and, i- i- uh, T_I-digits we have both
	</Segment>
	<Segment StartTime="1002.563" EndTime="1007.316" Participant="mn007">
	uh, digits <Pause/> training data and also <Pause/> more general training data. So.
	</Segment>
	<Segment StartTime="1008.918" EndTime="1009.602" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1009.656" EndTime="1016.510" Participant="me013">
	Well, we also have this Broadcast News that we were talking about taking off the disk, which is - <VocalSound Description="laugh"/> is microphone data for - for English.
	</Segment>
	<Segment StartTime="1014.204" EndTime="1014.989" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="1016.510" EndTime="1021.353" Participant="mn007">
	Yeah, perhaps - yeah, there is also TIMIT. We could use TIMIT.
	</Segment>
	<Segment StartTime="1019.495" EndTime="1020.010" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1020.331" EndTime="1020.795" Participant="me013">
	Right.
	</Segment>
	<Segment StartTime="1022.389" EndTime="1023.796" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="1023.528" EndTime="1025.298" Participant="me013">
	Yeah, so there's plenty of stuff around.
	</Segment>
	<Segment StartTime="1026.278" EndTime="1029.684" Participant="me013">
	O_K, so anyway, th- the basic plan is to, uh, test this cube.
	</Segment>
	<Segment StartTime="1030.542" EndTime="1031.066" Participant="me013">
	<Uncertain NumSyllables="1"> @@ </Uncertain>
	</Segment>
	<Segment StartTime="1030.945" EndTime="1033.002" Participant="mn007">
	Yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1031.066" EndTime="1035.222" Participant="me013">
	Yes. <VocalSound Description="laugh"/> To fill i- fill it in, yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1031.718" EndTime="1033.137" Participant="fn002">
	To fill in the cube. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="1034.575" EndTime="1036.303" Participant="mn007">
	<VocalSound Description="laugh"/> Uh.
	</Segment>
	<Segment StartTime="1035.222" EndTime="1036.757" Participant="me013">
	<VocalSound Description="inbreath"/> O_K.
	</Segment>
	<Segment StartTime="1037.439" EndTime="1040.303" Participant="mn007">
	Yeah, and perhaps, um - <Pause/>
	</Segment>
	<Segment StartTime="1040.303" EndTime="1051.372" Participant="mn007">
	We were thinking that perhaps the cross-language issue is not, uh, so big of a issue. Well, w- w- we - perhaps we should not focus too much on that cross-language stuff.
	</Segment>
	<Segment StartTime="1051.641" EndTime="1056.580" Participant="mn007">
	I mean, uh, training - training a net on a language and testing a- for another language.
	</Segment>
	<Segment StartTime="1052.304" EndTime="1053.012" Participant="me013">
	Uh-huh.
	</Segment>
	<Segment StartTime="1057.424" EndTime="1058.416" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1058.416" EndTime="1067.400" Participant="mn007">
	Perhaps the most important is to have neural networks trained on the target languages. But, uh, with a general database - general databases.
	</Segment>
	<Segment StartTime="1058.529" EndTime="1059.327" Participant="me013">
	But that's -
	</Segment>
	<Segment StartTime="1059.327" EndTime="1060.192" Participant="me013">
	<VocalSound Description="squeaks"/>
	</Segment>
	<Segment StartTime="1068.057" EndTime="1068.428" Participant="mn007">
	u-
	</Segment>
	<Segment StartTime="1068.600" EndTime="1070.145" Participant="mn007">
	So that th- Well,
	</Segment>
	<Segment StartTime="1070.289" EndTime="1076.336" Participant="mn007">
	the - the guy who has to develop an application with one language can use the net trained o- on that language,
	</Segment>
	<Segment StartTime="1076.336" EndTime="1078.495" Participant="mn007">
	or a generic net, but not trained on a -
	</Segment>
	<Segment StartTime="1076.449" EndTime="1083.980" Participant="me013">
	Uh, depen- it depen- it depends how you mean "using the net". So, if you're talking about for producing these discriminative features <Pause/> that we're talking about <Pause/> you <Emphasis> can't </Emphasis> do that.
	</Segment>
	<Segment StartTime="1081.305" EndTime="1081.752" Participant="mn007">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="1082.250" EndTime="1082.740" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1084.860" EndTime="1086.280" Participant="me013">
	Because - because
	</Segment>
	<Segment StartTime="1086.498" EndTime="1089.846" Participant="me013">
	the - what they're asking for is - is a feature set.
	</Segment>
	<Segment StartTime="1090.510" EndTime="1090.901" Participant="me013">
	Right?
	</Segment>
	<Segment StartTime="1090.510" EndTime="1090.901" Participant="mn007">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="1090.901" EndTime="1091.530" Participant="me013">
	And so,
	</Segment>
	<Segment StartTime="1092.020" EndTime="1096.318" Participant="me013">
	uh, we're the ones who have been weird by - by doing this training.
	</Segment>
	<Segment StartTime="1096.441" EndTime="1096.865" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="1096.621" EndTime="1101.446" Participant="me013">
	But if we say, "No, you have to have a different feature set for each language," I think this is ver- gonna be very bad.
	</Segment>
	<Segment StartTime="1101.329" EndTime="1102.111" Participant="me006">
	Oh.
	</Segment>
	<Segment StartTime="1101.967" EndTime="1103.022" Participant="mn007">
	You think so. Mmm.
	</Segment>
	<Segment StartTime="1102.043" EndTime="1104.462" Participant="me013">
	So - Oh yeah. Yeah. I mean,
	</Segment>
	<Segment StartTime="1102.187" EndTime="1102.680" Participant="fn002">
	Oh.
	</Segment>
	<Segment StartTime="1102.630" EndTime="1103.187" Participant="me006">
	That's -
	</Segment>
	<Segment StartTime="1104.965" EndTime="1109.054" Participant="me013">
	in <Emphasis> principle, </Emphasis> I mean <Emphasis> conceptually, </Emphasis> it's sort of like they want a re- <Uncertain NumSyllables="4"> @@ </Uncertain> <Comment Description="uncodeable vocal fragments"/>
	</Segment>
	<Segment StartTime="1106.033" EndTime="1106.561" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1109.054" EndTime="1110.781" Participant="me013">
	well, they want a replacement for mel cepstra.
	</Segment>
	<Segment StartTime="1110.781" EndTime="1117.116" Participant="me013">
	So, we say "O_K, this is the year two thousand, we've got something much better than mel cepstra. It's, you know, gobbledy-gook." O_K? And so
	</Segment>
	<Segment StartTime="1116.862" EndTime="1119.519" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1117.116" EndTime="1121.831" Participant="me013">
	<VocalSound Description="inbreath"/> we give them these gobbledy-gook features but these gobbledy-gook features are supposed to be good for any language.
	</Segment>
	<Segment StartTime="1121.894" EndTime="1122.267" Participant="mn007">
	Hmm.
	</Segment>
	<Segment StartTime="1122.630" EndTime="1128.834" Participant="me013">
	Cuz you don't know who's gonna call, and you know, I mean so it's - it's - it's, uh, uh - how do you know what language it is?
	</Segment>
	<Segment StartTime="1129.804" EndTime="1130.981" Participant="me013">
	Somebody picks up the phone.
	</Segment>
	<Segment StartTime="1131.394" EndTime="1132.440" Participant="me013">
	So thi- this is their image.
	</Segment>
	<Segment StartTime="1131.962" EndTime="1133.615" Participant="mn007">
	Well, I <Comment Description="creaky voice"/> <Pronounce Pronunciation="palatal fricative"> chh </Pronounce> - <Comment Description="nonword, vocal gesture of difficulty"/>
	</Segment>
	<Segment StartTime="1132.440" EndTime="1133.841" Participant="me013">
	Someone picks up the phone, right?
	</Segment>
	<Segment StartTime="1133.841" EndTime="1135.678" Participant="me013">
	And - and he - he picks up the ph-
	</Segment>
	<Segment StartTime="1135.007" EndTime="1139.774" Participant="mn007">
	Yeah, but the - the application is - there is a target language for the application. So, if a -
	</Segment>
	<Segment StartTime="1139.074" EndTime="1140.296" Participant="me013">
	Yeah. y- y- y- <Comment Description="creaky voice"/>
	</Segment>
	<Segment StartTime="1140.400" EndTime="1141.344" Participant="me013">
	Well.
	</Segment>
	<Segment StartTime="1140.833" EndTime="1141.344" Participant="mn007">
	Well.
	</Segment>
	<Segment StartTime="1141.344" EndTime="1143.539" Participant="me013">
	But, no but, y- you - you pick up the phone,
	</Segment>
	<Segment StartTime="1143.431" EndTime="1143.889" Participant="mn007">
	Yeah?
	</Segment>
	<Segment StartTime="1143.889" EndTime="1145.911" Participant="me013">
	you talk on the phone, and it sends features out.
	</Segment>
	<Segment StartTime="1146.657" EndTime="1148.931" Participant="me013">
	O_K, so the phone doesn't know
	</Segment>
	<Segment StartTime="1149.570" EndTime="1151.311" Participant="me013">
	what a - what - what your language is.
	</Segment>
	<Segment StartTime="1149.570" EndTime="1150.670" Participant="mn007">
	Yeah, if - Yeah.
	</Segment>
	<Segment StartTime="1151.600" EndTime="1157.214" Participant="mn007">
	If it's th- in the phone, but - well, it - that - that could be th- at the server's side, and, well.
	</Segment>
	<Segment StartTime="1153.471" EndTime="1155.327" Participant="me013">
	But that's the image that they <Emphasis> have. </Emphasis>
	</Segment>
	<Segment StartTime="1156.600" EndTime="1158.864" Participant="me013">
	It could be, but that's the <Emphasis> image </Emphasis> they have, right?
	</Segment>
	<Segment StartTime="1158.098" EndTime="1158.864" Participant="mn007">
	Mmm, yeah.
	</Segment>
	<Segment StartTime="1158.864" EndTime="1164.748" Participant="me013">
	So that's - that's - I mean, one could argue all over the place about how things really will be in ten years.
	</Segment>
	<Segment StartTime="1165.087" EndTime="1170.695" Participant="me013">
	But the particular image that the cellular industry has right now is that it's distributed speech recognition,
	</Segment>
	<Segment StartTime="1170.695" EndTime="1176.933" Participant="me013">
	where the, uh, uh, probabilistic part, and - and s- semantics and so forth are all on the servers,
	</Segment>
	<Segment StartTime="1176.933" EndTime="1179.203" Participant="me013">
	and you compute features of the - uh, on the phone.
	</Segment>
	<Segment StartTime="1179.671" EndTime="1184.985" Participant="me013">
	So that's - that's what we're involved in. We might - might or might not agree that that's the way it will be in ten years, but
	</Segment>
	<Segment StartTime="1184.985" EndTime="1187.476" Participant="me013">
	that's - that's - that's what they're asking for.
	</Segment>
	<Segment StartTime="1187.476" EndTime="1189.140" Participant="me013">
	So - so I think
	</Segment>
	<Segment StartTime="1189.140" EndTime="1192.977" Participant="me013">
	that - th- th- it is an important issue whether it works cross-language.
	</Segment>
	<Segment StartTime="1193.256" EndTime="1199.894" Participant="me013">
	Now, it's the O_G_I, uh, folks' perspective right now that probably that's not the biggest deal.
	</Segment>
	<Segment StartTime="1199.894" EndTime="1202.930" Participant="me013">
	And that the biggest deal is the, um
	</Segment>
	<Segment StartTime="1202.930" EndTime="1205.444" Participant="me013">
	envir- acoustic-environment mismatch.
	</Segment>
	<Segment StartTime="1205.444" EndTime="1210.965" Participant="me013">
	And they may very well be right, but I - I was hoping we could just do a test and determine if that was true.
	</Segment>
	<Segment StartTime="1211.271" EndTime="1217.504" Participant="me013">
	If that's true, we don't need to worry so much. Maybe - maybe we have a couple languages in the training set and that gives us enough breadth
	</Segment>
	<Segment StartTime="1217.940" EndTime="1220.929" Participant="me013">
	uh, uh, that - that - that the rest doesn't matter.
	</Segment>
	<Segment StartTime="1221.328" EndTime="1225.018" Participant="me013">
	Um, the other thing is, uh,
	</Segment>
	<Segment StartTime="1225.462" EndTime="1227.960" Participant="me013">
	this notion of training to uh -
	</Segment>
	<Segment StartTime="1227.960" EndTime="1229.713" Participant="me013">
	which I - I guess they're starting to look at up there, <Comment Description="embedded"/>
	</Segment>
	<Segment StartTime="1229.713" EndTime="1233.226" Participant="me013">
	training to something more like articulatory features.
	</Segment>
	<Segment StartTime="1233.226" EndTime="1240.884" Participant="me013">
	Uh, and if you have something that's just good for distinguishing different articulatory features that should just be good across, you know, a wide range of languages.
	</Segment>
	<Segment StartTime="1236.560" EndTime="1237.255" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1240.331" EndTime="1241.049" Participant="mn007">
	Yeah. Yeah.
	</Segment>
	<Segment StartTime="1240.884" EndTime="1242.340" Participant="me013">
	Uh, but -
	</Segment>
	<Segment StartTime="1242.340" EndTime="1246.778" Participant="me013">
	Yeah, so I don't th- I know - unfortunately I don't - I see what you're comi- where you're coming from, I think, but I don't think we can ignore it.
	</Segment>
	<Segment StartTime="1247.511" EndTime="1257.670" Participant="mn007">
	So we - we really have to do test with a real cross-language. I mean, tr- for instance training on English and testing on Italian, or -
	</Segment>
	<Segment StartTime="1257.670" EndTime="1258.801" Participant="mn007">
	Or we can train -
	</Segment>
	<Segment StartTime="1259.628" EndTime="1265.306" Participant="mn007">
	or else, uh, can we train a net on, uh, a range of languages and -
	</Segment>
	<Segment StartTime="1266.453" EndTime="1267.912" Participant="me006">
	<Emphasis> Test </Emphasis> on an <Emphasis> unseen. </Emphasis>
	</Segment>
	<Segment StartTime="1266.618" EndTime="1270.329" Participant="mn007">
	which can include the test - the test <Uncertain NumSyllables="1"> @@ </Uncertain> the target language, or -
	</Segment>
	<Segment StartTime="1270.329" EndTime="1271.772" Participant="me013">
	Yeah, so, um,
	</Segment>
	<Segment StartTime="1272.090" EndTime="1276.787" Participant="me013">
	there's - there's, uh - This is complex. So, ultimately,
	</Segment>
	<Segment StartTime="1277.410" EndTime="1282.764" Participant="me013">
	uh, as I was saying, I think it doesn't fit within their image that you switch nets based on language.
	</Segment>
	<Segment StartTime="1282.654" EndTime="1283.175" Participant="mn007">
	<VocalSound Description="mouth"/> Yeah.
	</Segment>
	<Segment StartTime="1283.175" EndTime="1283.590" Participant="me013">
	Now,
	</Segment>
	<Segment StartTime="1284.070" EndTime="1287.582" Participant="me013">
	can you include, uh, the - the target language?
	</Segment>
	<Segment StartTime="1288.160" EndTime="1288.678" Participant="me013">
	Um,
	</Segment>
	<Segment StartTime="1289.656" EndTime="1292.606" Participant="me013">
	from a purist's standpoint it'd be nice not to
	</Segment>
	<Segment StartTime="1292.993" EndTime="1295.280" Participant="me013">
	because then you can say when -
	</Segment>
	<Segment StartTime="1295.280" EndTime="1307.307" Participant="me013">
	because surely someone is going to say at some point, "O_K, so you put in the German and the Finnish. Uh, now, what do you do, uh, when somebody has Portuguese?" you know?
	</Segment>
	<Segment StartTime="1298.081" EndTime="1298.818" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1307.307" EndTime="1308.872" Participant="me013">
	Um, and -
	</Segment>
	<Segment StartTime="1309.200" EndTime="1310.672" Participant="me013">
	Uh, <Emphasis> however, </Emphasis>
	</Segment>
	<Segment StartTime="1311.760" EndTime="1314.480" Participant="me013">
	you aren't - it isn't actually a constraint
	</Segment>
	<Segment StartTime="1314.480" EndTime="1316.179" Participant="me013">
	in this evaluation.
	</Segment>
	<Segment StartTime="1316.179" EndTime="1317.133" Participant="me013">
	So I would say
	</Segment>
	<Segment StartTime="1317.420" EndTime="1318.857" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1317.500" EndTime="1320.774" Participant="me013">
	if it looks like there's a big difference to put it in,
	</Segment>
	<Segment StartTime="1320.774" EndTime="1327.605" Participant="me013">
	then we'd make note of it, and then we probably put in the other, because we have so many other problems in trying to get things to work well here that -
	</Segment>
	<Segment StartTime="1327.157" EndTime="1327.697" Participant="mn007">
	Mmm?
	</Segment>
	<Segment StartTime="1327.605" EndTime="1332.716" Participant="me013">
	that, you know, it's not so bad as long as we - we note it and say, "Look, we <Emphasis> did </Emphasis> do this".
	</Segment>
	<Segment StartTime="1332.505" EndTime="1335.235" Participant="me018">
	And so, ideally, what you'd wanna do is you'd wanna
	</Segment>
	<Segment StartTime="1332.716" EndTime="1333.343" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="1335.235" EndTime="1340.903" Participant="me018">
	run it with and without the target language and the training set for a wide range of languages.
	</Segment>
	<Segment StartTime="1338.406" EndTime="1338.890" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1338.535" EndTime="1339.515" Participant="mn007">
	Yeah, perhaps. Yeah.
	</Segment>
	<Segment StartTime="1340.765" EndTime="1341.293" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="1340.903" EndTime="1342.244" Participant="me018">
	And that way you can say, "Well,"
	</Segment>
	<Segment StartTime="1340.903" EndTime="1341.524" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1342.700" EndTime="1343.173" Participant="me018">
	you know,
	</Segment>
	<Segment StartTime="1343.337" EndTime="1347.402" Participant="me018">
	"we're gonna build it for what we <Emphasis> think </Emphasis> are <Pause/> the most <Emphasis> common </Emphasis> ones", but <Pronounce Pronunciation=""ifff""> if </Pronounce>
	</Segment>
	<Segment StartTime="1347.402" EndTime="1349.543" Participant="me018">
	that - somebody uses it with a different language,
	</Segment>
	<Segment StartTime="1349.884" EndTime="1352.431" Participant="me018">
	you know, "here's what's you're l- here's what's likely to happen."
	</Segment>
	<Segment StartTime="1352.805" EndTime="1358.400" Participant="me013">
	Yeah, cuz the truth is, is that it's - it's not like there are - I mean, al- although there are thousands of languages,
	</Segment>
	<Segment StartTime="1358.400" EndTime="1365.079" Participant="me013">
	uh, from uh, uh, the point of view of cellular companies, there aren't. There's - <VocalSound Description="laugh"/> you know, there's fifty or something, you know? So,
	</Segment>
	<Segment StartTime="1361.752" EndTime="1362.187" Participant="me018">
	Right.
	</Segment>
	<Segment StartTime="1362.187" EndTime="1362.944" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1365.079" EndTime="1367.026" Participant="me013">
	uh, an- and they aren't -
	</Segment>
	<Segment StartTime="1367.026" EndTime="1367.579" Participant="me013">
	you know,
	</Segment>
	<Segment StartTime="1367.579" EndTime="1371.602" Participant="me013">
	with the exception of Finnish, which I guess it's pretty different from most - most things.
	</Segment>
	<Segment StartTime="1371.117" EndTime="1371.989" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1371.602" EndTime="1375.092" Participant="me013">
	uh, it's - it's, uh - most of them are
	</Segment>
	<Segment StartTime="1375.470" EndTime="1377.083" Participant="me013">
	like at least some of the others.
	</Segment>
	<Segment StartTime="1377.083" EndTime="1379.963" Participant="me013">
	And so, our guess that Spanish is like Italian, and -
	</Segment>
	<Segment StartTime="1381.056" EndTime="1382.308" Participant="me013">
	and so on.
	</Segment>
	<Segment StartTime="1382.970" EndTime="1386.992" Participant="me013">
	I guess Finnish is a - is - is a little bit like Hungarian, supposedly, right?
	</Segment>
	<Segment StartTime="1386.992" EndTime="1389.448" Participant="me013">
	Or is - I think - well, I kn- oh, well I know that
	</Segment>
	<Segment StartTime="1387.348" EndTime="1388.605" Participant="me018">
	I don't know anything about Finnish.
	</Segment>
	<Segment StartTime="1389.448" EndTime="1395.878" Participant="me013">
	H- uh, H- I mean, I'm not a linguist, but I guess Hungarian and Finnish and one of the - one of the languages from the former Soviet Union
	</Segment>
	<Segment StartTime="1396.332" EndTime="1398.091" Participant="me013">
	are in this sort of same family.
	</Segment>
	<Segment StartTime="1398.091" EndTime="1399.242" Participant="me013">
	But they're just these, you know,
	</Segment>
	<Segment StartTime="1398.223" EndTime="1398.578" Participant="me018">
	Hmm.
	</Segment>
	<Segment StartTime="1399.787" EndTime="1402.445" Participant="me013">
	uh - countries that are pretty far apart from one another, have -
	</Segment>
	<Segment StartTime="1401.915" EndTime="1402.336" Participant="me018">
	Hmm.
	</Segment>
	<Segment StartTime="1402.070" EndTime="1402.679" Participant="mn007">
	Mmm. <Comment Description="quietly"/>
	</Segment>
	<Segment StartTime="1402.445" EndTime="1403.238" Participant="me013">
	I guess,
	</Segment>
	<Segment StartTime="1404.057" EndTime="1406.214" Participant="me013">
	people rode in on horses and
	</Segment>
	<Segment StartTime="1405.564" EndTime="1406.629" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1406.214" EndTime="1406.713" Participant="me013">
	brought their -
	</Segment>
	<Segment StartTime="1408.497" EndTime="1409.230" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="1412.493" EndTime="1414.144" Participant="mn007">
	The - Yeah.
	</Segment>
	<Segment StartTime="1413.935" EndTime="1414.500" Participant="me013">
	Your turn.
	</Segment>
	<Segment StartTime="1414.403" EndTime="1417.123" Participant="me006">
	Oh, my turn. Oh, O_K. Um,
	</Segment>
	<Segment StartTime="1414.500" EndTime="1416.691" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1417.123" EndTime="1421.663" Participant="me006">
	Let's see, I - I spent the last week, uh, looking over Stephane's shoulder.
	</Segment>
	<Segment StartTime="1421.830" EndTime="1422.578" Participant="mn007">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="1422.324" EndTime="1426.141" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1422.324" EndTime="1426.203" Participant="me006">
	And - <VocalSound Description="laugh"/> and understanding some of the data.
	</Segment>
	<Segment StartTime="1422.578" EndTime="1424.931" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1426.203" EndTime="1427.542" Participant="me006">
	I re-installed, um,
	</Segment>
	<Segment StartTime="1428.039" EndTime="1430.389" Participant="me006">
	um, H_T_K, the free version,
	</Segment>
	<Segment StartTime="1430.902" EndTime="1437.488" Participant="me006">
	so, um, everybody's now using three point O_, which is the same version that, uh, O_G_I is using.
	</Segment>
	<Segment StartTime="1437.755" EndTime="1438.284" Participant="me013">
	Oh, good.
	</Segment>
	<Segment StartTime="1438.284" EndTime="1443.460" Participant="me006">
	Yeah. So, without - without any licensing big deals, or anything like that.
	</Segment>
	<Segment StartTime="1442.986" EndTime="1443.666" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1443.460" EndTime="1449.081" Participant="me006">
	And, um, so we've been talking about this - this, uh, cube thing,
	</Segment>
	<Segment StartTime="1449.081" EndTime="1454.252" Participant="me006">
	and it's beginning more and more looking like the, uh, the Borge cube thing.
	</Segment>
	<Segment StartTime="1454.252" EndTime="1455.944" Participant="me006">
	It's really gargantuan.
	</Segment>
	<Segment StartTime="1455.690" EndTime="1457.045" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1455.944" EndTime="1459.192" Participant="me006">
	Um, <VocalSound Description="mouth"/> but I-
	</Segment>
	<Segment StartTime="1457.679" EndTime="1459.062" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1459.183" EndTime="1460.647" Participant="me013">
	So are - are you going to be assimilated?
	</Segment>
	<Segment StartTime="1459.192" EndTime="1460.029" Participant="me006">
	I'm -
	</Segment>
	<Segment StartTime="1460.708" EndTime="1462.249" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1460.881" EndTime="1464.277" Participant="me006">
	Am I - <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1460.891" EndTime="1464.444" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1462.249" EndTime="1465.234" Participant="me018">
	<VocalSound Description="laugh"/> Resistance is futile. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1464.277" EndTime="1466.278" Participant="me006">
	Exactly. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1466.278" EndTime="1470.367" Participant="me006">
	Um, yeah, so I- I've been looking at, uh, uh, TIMIT stuff. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1466.989" EndTime="1467.604" Participant="mn007">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="1470.874" EndTime="1475.109" Participant="me006">
	Um, the - the stuff that we've been working on with TIMIT, trying to get a,
	</Segment>
	<Segment StartTime="1475.109" EndTime="1487.646" Participant="me006">
	um - a labels file so we can, uh, train up a - train up a net on TIMIT and test, um, the difference between this net trained on TIMIT and a net trained on digits alone.
	</Segment>
	<Segment StartTime="1487.646" EndTime="1488.505" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1487.646" EndTime="1492.256" Participant="me006">
	Um, and seeing if - if it hurts or helps.
	</Segment>
	<Segment StartTime="1492.256" EndTime="1497.855" Participant="me013">
	And again, when y- just to clarify, when you're talking about training up a net, you're talking about training up a net for a tandem
	</Segment>
	<Segment StartTime="1492.256" EndTime="1493.193" Participant="me006">
	Anyway.
	</Segment>
	<Segment StartTime="1497.837" EndTime="1499.392" Participant="me006">
	Yeah, yeah. Um.
	</Segment>
	<Segment StartTime="1498.621" EndTime="1504.145" Participant="me013">
	approach? And - and the inputs are P_L_P and delta and that sort of thing, or - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="1499.392" EndTime="1499.868" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1502.623" EndTime="1517.445" Participant="me006">
	Well, the inputs are one dimension of the cube, which, um, we've talked about it being, uh, P_L_P, um, M_F_C_Cs, um, J_- J_RASTA, J_RASTA-L_D_A - <Comment Description="list"/>
	</Segment>
	<Segment StartTime="1508.690" EndTime="1509.457" Participant="mn007">
	Hmm. <Comment Description="quietly"/>
	</Segment>
	<Segment StartTime="1517.571" EndTime="1521.743" Participant="me013">
	Yeah, but your initial things you're making one choice there, right? Which is P_L_P, or something? Yeah.
	</Segment>
	<Segment StartTime="1519.543" EndTime="1521.029" Participant="me006">
	Yeah, right.
	</Segment>
	<Segment StartTime="1521.361" EndTime="1527.420" Participant="me006">
	Um, I - I haven't - I haven't decided on - on the initial thing. Probably - probably something like P_L_P.
	</Segment>
	<Segment StartTime="1527.462" EndTime="1528.144" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1527.669" EndTime="1528.452" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="1528.399" EndTime="1528.932" Participant="mn007">
	Hmm.
	</Segment>
	<Segment StartTime="1529.063" EndTime="1541.666" Participant="me013">
	Um, so - so you take P_L_P and you - you, uh, do it - uh, you - you, uh, use H_T_K with it with the transformed features using a neural net that's trained.
	</Segment>
	<Segment StartTime="1541.666" EndTime="1552.600" Participant="me013">
	And the training could either be from Digits itself or from TIMIT. And that's the - and, and th- and then the testing would be these other things which - which - which might be foreign language.
	</Segment>
	<Segment StartTime="1546.241" EndTime="1546.901" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="1552.866" EndTime="1553.358" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="1553.358" EndTime="1554.138" Participant="me013">
	I see.
	</Segment>
	<Segment StartTime="1553.358" EndTime="1554.138" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="1554.138" EndTime="1556.755" Participant="me013">
	I - I - I get in the picture about the cube. O_K.
	</Segment>
	<Segment StartTime="1555.698" EndTime="1556.433" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="1556.433" EndTime="1557.749" Participant="me006">
	<Uncertain> Maybe - </Uncertain> O_K. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="1556.693" EndTime="1558.033" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1557.277" EndTime="1557.636" Participant="me013">
	O_K. <Comment Description="while smiling"/>
	</Segment>
	<Segment StartTime="1557.636" EndTime="1562.725" Participant="me013">
	Um, I mean, those listening to this will not have a picture either, so, um, I guess I'm - I'm not any worse off.
	</Segment>
	<Segment StartTime="1557.749" EndTime="1558.174" Participant="me006">
	Uh-huh.
	</Segment>
	<Segment StartTime="1558.174" EndTime="1558.924" Participant="me006">
	<VocalSound Description="inbreath"/> <Comment Description="as if to speak"/>
	</Segment>
	<Segment StartTime="1562.725" EndTime="1569.985" Participant="me013">
	But <VocalSound Description="laugh"/> but at some point - somebody should just show me the cube. It sounds s- I - I get - I think I get the general idea of it, yeah.
	</Segment>
	<Segment StartTime="1563.024" EndTime="1567.936" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1568.532" EndTime="1569.985" Participant="me006">
	Yeah, yeah, b-
	</Segment>
	<Segment StartTime="1569.050" EndTime="1573.361" Participant="mn007">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="1569.985" EndTime="1570.351" Participant="me006">
	May-
	</Segment>
	<Segment StartTime="1569.985" EndTime="1572.244" Participant="me018">
	So, when you said that you were getting the labels for TIMIT, <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1572.204" EndTime="1572.715" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1572.786" EndTime="1575.266" Participant="me018">
	um, are y- what do you mean by that?
	</Segment>
	<Segment StartTime="1574.634" EndTime="1585.125" Participant="me006">
	Oh, I'm just - I'm just, uh, transforming them from the, um, the standard TIMIT transcriptions into - into a nice long huge P_file to do training.
	</Segment>
	<Segment StartTime="1580.705" EndTime="1581.294" Participant="me018">
	Mmm.
	</Segment>
	<Segment StartTime="1583.853" EndTime="1587.189" Participant="me018">
	Were the digits, um, hand-labeled for phones?
	</Segment>
	<Segment StartTime="1587.509" EndTime="1589.834" Participant="me006">
	Um, the - the digits - <Comment Description="considering"/>
	</Segment>
	<Segment StartTime="1588.095" EndTime="1590.611" Participant="me018">
	Or were they - those labels automatically derived?
	</Segment>
	<Segment StartTime="1590.528" EndTime="1595.974" Participant="me006">
	Oh yeah, those were - those were automatically derived by - by Dan
	</Segment>
	<Segment StartTime="1595.609" EndTime="1596.113" Participant="me018">
	Mmm.
	</Segment>
	<Segment StartTime="1595.974" EndTime="1599.575" Participant="me006">
	using, um, embedded - embedded training and alignment.
	</Segment>
	<Segment StartTime="1599.575" EndTime="1600.847" Participant="me013">
	Ah, but which Dan?
	</Segment>
	<Segment StartTime="1600.847" EndTime="1602.706" Participant="me006">
	Uh, <Emphasis> Ellis. </Emphasis> Right?
	</Segment>
	<Segment StartTime="1602.160" EndTime="1602.706" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="1603.182" EndTime="1603.962" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="1603.995" EndTime="1605.113" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1605.197" EndTime="1611.133" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1605.197" EndTime="1605.667" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="1605.667" EndTime="1606.634" Participant="me006">
	So.
	</Segment>
	<Segment StartTime="1606.634" EndTime="1610.397" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1608.290" EndTime="1615.216" Participant="me018">
	I was just wondering because that test you're t- I - I think you're doing this test because you want to determine whether or not,
	</Segment>
	<Segment StartTime="1610.397" EndTime="1611.034" Participant="me006">
	Uh-huh.
	</Segment>
	<Segment StartTime="1615.216" EndTime="1620.805" Participant="me018">
	uh, having s- general speech performs as well as having specific <Pause/> speech.
	</Segment>
	<Segment StartTime="1620.469" EndTime="1621.326" Participant="me006">
	That's right.
	</Segment>
	<Segment StartTime="1621.326" EndTime="1629.110" Participant="me013">
	Well, especially when you go over the different languages again, because you'd - the different languages have different words for the different digits, so it's -
	</Segment>
	<Segment StartTime="1627.329" EndTime="1637.988" Participant="me018">
	Mm-hmm. And I was - yeah, so I was just wondering if the fact that TIMIT - you're using the hand-labeled stuff from TIMIT might be - confuse the results that you get.
	</Segment>
	<Segment StartTime="1637.988" EndTime="1641.540" Participant="me013">
	I - I think it would, but - but on the other hand it might be better.
	</Segment>
	<Segment StartTime="1642.834" EndTime="1644.690" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1642.897" EndTime="1646.524" Participant="me018">
	Right, but if it's better, it may be better because <Pause/> it was hand-labeled.
	</Segment>
	<Segment StartTime="1645.105" EndTime="1647.096" Participant="me013">
	Oh yeah, but still <Uncertain NumSyllables="1"> @@ </Uncertain> probably use it.
	</Segment>
	<Segment StartTime="1647.509" EndTime="1647.983" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="1648.022" EndTime="1652.304" Participant="me013">
	I mean, you know, I - I - I guess I'm sounding cavalier, but I mean, I think
	</Segment>
	<Segment StartTime="1648.185" EndTime="1648.701" Participant="me018">
	O_K.
	</Segment>
	<Segment StartTime="1652.304" EndTime="1658.381" Participant="me013">
	the point is you have, uh, a bunch of labels and - and they're han- hand uh - hand-marked.
	</Segment>
	<Segment StartTime="1658.381" EndTime="1663.761" Participant="me013">
	Uh, I guess, actually, TIMIT was not entirely hand-marked. It was automatically first, and then hand - hand-corrected.
	</Segment>
	<Segment StartTime="1663.761" EndTime="1664.947" Participant="me018">
	Oh, O_K.
	</Segment>
	<Segment StartTime="1663.761" EndTime="1672.092" Participant="me013">
	But - but, um, uh, it - it, um, it might be a better source.
	</Segment>
	<Segment StartTime="1664.947" EndTime="1666.509" Participant="me018">
	<VocalSound Description="cough"/>
	</Segment>
	<Segment StartTime="1672.092" EndTime="1679.693" Participant="me013">
	So, i- it's - you're right. It would be another interesting scientific question to ask, "Is it because it's a broad source or because it was, you know, carefully?" uh.
	</Segment>
	<Segment StartTime="1679.472" EndTime="1680.060" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1680.300" EndTime="1690.378" Participant="me013">
	And that's something you could ask, but given limited time, I think the main thing is if it's a better thing for going across languages on this training tandem system, then it's probably -
	</Segment>
	<Segment StartTime="1682.444" EndTime="1683.153" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="1685.034" EndTime="1685.701" Participant="mn007">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="1688.203" EndTime="1688.765" Participant="me018">
	Right.
	</Segment>
	<Segment StartTime="1689.759" EndTime="1691.875" Participant="me018">
	What about the differences in the phone sets?
	</Segment>
	<Segment StartTime="1692.653" EndTime="1694.171" Participant="me006">
	Uh, between languages?
	</Segment>
	<Segment StartTime="1694.171" EndTime="1697.916" Participant="me018">
	No, between TIMIT and the - the digits.
	</Segment>
	<Segment StartTime="1697.227" EndTime="1699.342" Participant="me006">
	Oh, um, right.
	</Segment>
	<Segment StartTime="1699.342" EndTime="1704.467" Participant="me006">
	Well, there's a mapping from the sixty-one phonemes in TIMIT to -
	</Segment>
	<Segment StartTime="1703.592" EndTime="1704.537" Participant="fn002">
	Sixty-one. <Comment Description="whispered"/>
	</Segment>
	<Segment StartTime="1704.467" EndTime="1707.197" Participant="me006">
	to fifty-six, the ICSI fifty-six. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1705.766" EndTime="1707.233" Participant="me018">
	Oh, O_K. I see.
	</Segment>
	<Segment StartTime="1707.197" EndTime="1714.611" Participant="me006">
	And then the digits phonemes, um, there's about twenty- twenty-two or twenty-four of them? Is that right?
	</Segment>
	<Segment StartTime="1714.554" EndTime="1715.473" Participant="mn007">
	Yep.
	</Segment>
	<Segment StartTime="1714.568" EndTime="1716.700" Participant="me018">
	Out of that fifty-six? Oh, O_K.
	</Segment>
	<Segment StartTime="1715.354" EndTime="1716.700" Participant="me006">
	Out of that fifty-six.
	</Segment>
	<Segment StartTime="1717.071" EndTime="1717.432" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="1717.851" EndTime="1722.493" Participant="me006">
	So, it's - it's definitely broader, yeah.
	</Segment>
	<Segment StartTime="1725.830" EndTime="1730.504" Participant="mn007">
	But, actually, the issue of phoneti- phon- uh phone- phoneme mappings
	</Segment>
	<Segment StartTime="1730.504" EndTime="1735.151" Participant="mn007">
	will arise when we will do severa- use several languages because
	</Segment>
	<Segment StartTime="1730.542" EndTime="1731.386" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="1735.151" EndTime="1737.587" Participant="mn007">
	you - Well,
	</Segment>
	<Segment StartTime="1737.587" EndTime="1741.685" Participant="mn007">
	some phonemes are not, uh, in every languages, and -
	</Segment>
	<Segment StartTime="1741.685" EndTime="1745.720" Participant="mn007">
	So we plan to develop a subset of the phonemes,
	</Segment>
	<Segment StartTime="1745.720" EndTime="1750.352" Participant="mn007">
	uh, that includes, uh, all the phonemes of our training languages, and
	</Segment>
	<Segment StartTime="1750.322" EndTime="1750.902" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1750.588" EndTime="1751.257" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1751.405" EndTime="1755.575" Participant="mn007">
	use a network with kind of one hundred outputs or something like that.
	</Segment>
	<Segment StartTime="1755.889" EndTime="1757.866" Participant="me013">
	You mean a superset, sort of.
	</Segment>
	<Segment StartTime="1757.191" EndTime="1758.949" Participant="mn007">
	Uh, yeah, superset, yeah.
	</Segment>
	<Segment StartTime="1757.866" EndTime="1759.998" Participant="me013">
	Yeah. <VocalSound Description="laugh"/> Yeah.
	</Segment>
	<Segment StartTime="1758.412" EndTime="1759.831" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="1759.831" EndTime="1765.406" Participant="fn002">
	I th- I looks the SAMPA- SAMPA phone. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1759.865" EndTime="1760.246" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="1766.110" EndTime="1766.479" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="1766.479" EndTime="1776.287" Participant="fn002">
	SAMPA phone? For English - uh American English, and the - the - the language who have more phone are the English.
	</Segment>
	<Segment StartTime="1776.287" EndTime="1777.190" Participant="me018">
	<Emphasis> Mmm. </Emphasis>
	</Segment>
	<Segment StartTime="1777.190" EndTime="1779.735" Participant="fn002">
	Of the - these language.
	</Segment>
	<Segment StartTime="1779.735" EndTime="1790.051" Participant="fn002">
	But n- for example, in Spain, the Spanish have several phone that d- doesn't appear in the E- English and we thought to complete.
	</Segment>
	<Segment StartTime="1790.051" EndTime="1801.937" Participant="fn002">
	But for that, it needs - we must r- h- do a lot of work <VocalSound Description="laugh"/> because we need to generate new tran- transcription for the database that we have.
	</Segment>
	<Segment StartTime="1795.680" EndTime="1796.411" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1802.645" EndTime="1803.510" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1803.510" EndTime="1806.722" Participant="me034">
	Other than the language, is there a reason <Emphasis> not </Emphasis> to use the TIMIT phone set?
	</Segment>
	<Segment StartTime="1808.048" EndTime="1808.830" Participant="me034">
	Cuz it's larger?
	</Segment>
	<Segment StartTime="1809.818" EndTime="1813.014" Participant="me034">
	As opposed to the ICSI <Pause/> phone set?
	</Segment>
	<Segment StartTime="1813.014" EndTime="1816.488" Participant="me006">
	Oh, you mean why map the sixty-one to the fifty-six?
	</Segment>
	<Segment StartTime="1816.420" EndTime="1816.949" Participant="me034">
	Yeah.
	</Segment>
	<Segment StartTime="1817.014" EndTime="1818.263" Participant="me006">
	I don't know. I have -
	</Segment>
	<Segment StartTime="1818.263" EndTime="1825.139" Participant="me013">
	Um, I forget if that happened starting with you, or was it - o- or if it was Eric, afterwards who did that. But I think,
	</Segment>
	<Segment StartTime="1825.139" EndTime="1828.702" Participant="me013">
	basically, there were several of the phones that were just hardly ever there.
	</Segment>
	<Segment StartTime="1828.933" EndTime="1832.106" Participant="me018">
	Yeah, and I think some of them, they were making distinctions between
	</Segment>
	<Segment StartTime="1832.106" EndTime="1834.679" Participant="me018">
	silence at the end and silence at the beginning,
	</Segment>
	<Segment StartTime="1834.679" EndTime="1838.042" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1834.679" EndTime="1836.645" Participant="me018">
	when really they're <Pause/> both silence.
	</Segment>
	<Segment StartTime="1834.679" EndTime="1835.622" Participant="me034">
	Oh. <Comment Description="lengthened, creaky voice"/> <Comment Description="vocal gesture of difficulty"/>
	</Segment>
	<Segment StartTime="1836.645" EndTime="1840.079" Participant="me018">
	I th- I think it was things like that that got it mapped down to fifty-six.
	</Segment>
	<Segment StartTime="1836.811" EndTime="1837.265" Participant="me034">
	O_K.
	</Segment>
	<Segment StartTime="1838.991" EndTime="1846.912" Participant="me013">
	Yeah, especially in a system like ours, which is a discriminative system. You know, you're really asking this net to learn. <Uncertain NumSyllables="4"> @@ </Uncertain> <VocalSound Description="laugh"/> It's - it's kind of hard. So.
	</Segment>
	<Segment StartTime="1843.860" EndTime="1844.206" Participant="me034">
	Yeah.
	</Segment>
	<Segment StartTime="1844.407" EndTime="1845.082" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="1846.211" EndTime="1847.629" Participant="me018">
	There's not much difference, really.
	</Segment>
	<Segment StartTime="1847.629" EndTime="1849.882" Participant="me018">
	And <Pause/> the ones that are gone, I think are -
	</Segment>
	<Segment StartTime="1849.882" EndTime="1853.419" Participant="me018">
	I think there was - they also in TIMIT had like a glottal stop, which
	</Segment>
	<Segment StartTime="1853.287" EndTime="1853.806" Participant="me034">
	Mm-hmm.
	</Segment>
	<Segment StartTime="1853.419" EndTime="1855.976" Participant="me018">
	<Uncertain> was </Uncertain> basically a short period of silence, and
	</Segment>
	<Segment StartTime="1855.976" EndTime="1856.792" Participant="me018">
	so.
	</Segment>
	<Segment StartTime="1856.202" EndTime="1857.364" Participant="me034">
	Well, we have that now, too, right?
	</Segment>
	<Segment StartTime="1857.497" EndTime="1858.020" Participant="me018">
	I don't know.
	</Segment>
	<Segment StartTime="1858.072" EndTime="1858.925" Participant="me034">
	Yeah.
	</Segment>
	<Segment StartTime="1858.338" EndTime="1865.681" Participant="me013">
	i- It's actually pretty common that a lot of the recognition systems people use have things like - like, say thirty-nine,
	</Segment>
	<Segment StartTime="1858.742" EndTime="1859.388" Participant="me018">
	So. <Comment Description="creaky voice"/>
	</Segment>
	<Segment StartTime="1866.414" EndTime="1872.547" Participant="me013">
	phone symbols, right? Uh, and then they get the variety by - by bringing in the context, the phonetic context. Uh.
	</Segment>
	<Segment StartTime="1873.079" EndTime="1877.037" Participant="me013">
	So we actually have an unusually large number in - in what we tend to use here.
	</Segment>
	<Segment StartTime="1878.633" EndTime="1879.453" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="1880.122" EndTime="1885.067" Participant="me013">
	So, a- a- actually - maybe - now you've got me sort of intrigued. What - there's -
	</Segment>
	<Segment StartTime="1886.031" EndTime="1888.758" Participant="me013">
	Can you describe what - what's <Emphasis> on </Emphasis> the cube? I mean - <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1887.581" EndTime="1896.185" Participant="me006">
	Yeah, w- I th- I think that's a good idea to - to talk about the whole cube and maybe we could sections in the cube for people to work on.
	</Segment>
	<Segment StartTime="1887.852" EndTime="1889.337" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1889.484" EndTime="1890.686" Participant="me013">
	Yeah, yeah.
	</Segment>
	<Segment StartTime="1890.087" EndTime="1890.812" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="1892.139" EndTime="1892.775" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="1894.983" EndTime="1895.555" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1896.178" EndTime="1896.713" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="1896.978" EndTime="1897.920" Participant="me006">
	Um,
	</Segment>
	<Segment StartTime="1897.920" EndTime="1898.574" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="1898.790" EndTime="1905.537" Participant="me013">
	O_K, so even - even though the meeting recorder doesn't - doesn't, uh - and since you're not running a video camera we won't get this, but if you use a board it'll help us anyway.
	</Segment>
	<Segment StartTime="1898.790" EndTime="1900.181" Participant="me006">
	Uh, do you wanna do it?
	</Segment>
	<Segment StartTime="1905.537" EndTime="1911.825" Participant="me013">
	Uh, point out one of the limitations of this <VocalSound Description="laugh"/> medium, but you've got the wireless on, right? Yeah, so you can walk around.
	</Segment>
	<Segment StartTime="1905.637" EndTime="1906.444" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1906.078" EndTime="1906.933" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="1908.305" EndTime="1910.323" Participant="me006">
	O_K. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="1910.323" EndTime="1911.994" Participant="me006">
	Yeah, I have the wireless.
	</Segment>
	<Segment StartTime="1911.825" EndTime="1913.467" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1911.994" EndTime="1912.853" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="1912.853" EndTime="1916.061" Participant="me006">
	Can y- can <Emphasis> you </Emphasis> walk around too? No. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1916.061" EndTime="1918.876" Participant="me006">
	O_K, well, um, s-
	</Segment>
	<Segment StartTime="1916.243" EndTime="1920.051" Participant="me013">
	Uh, he can't, actually, but - <VocalSound Description="laugh"/> He's tethered.
	</Segment>
	<Segment StartTime="1918.033" EndTime="1918.711" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="1918.876" EndTime="1922.702" Participant="me006">
	basically, the - the cube will have three dimensions.
	</Segment>
	<Segment StartTime="1922.702" EndTime="1926.686" Participant="me006">
	The first dimension is the - the features that we're going to use.
	</Segment>
	<Segment StartTime="1928.283" EndTime="1934.554" Participant="me006">
	And the second dimension, um, is the training corpus. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1935.506" EndTime="1938.895" Participant="me006">
	And that's the training on the discriminant neural net.
	</Segment>
	<Segment StartTime="1938.895" EndTime="1942.648" Participant="me006">
	Um <Pause/> and <Pause/> <NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="1942.648" EndTime="1944.554" Participant="me006">
	the <Emphasis> last </Emphasis> dimension
	</Segment>
	<Segment StartTime="1944.513" EndTime="1955.028" Participant="me013">
	Yeah and again - Yeah. So the - the training for H_T_K is always - that's always set up for the individual test, right? That there's some training data and some test data. So that's different than this. Yeah.
	</Segment>
	<Segment StartTime="1944.554" EndTime="1945.543" Participant="me006">
	happens to be -
	</Segment>
	<Segment StartTime="1953.969" EndTime="1957.126" Participant="me006">
	Right, right. This is - this is for -
	</Segment>
	<Segment StartTime="1957.126" EndTime="1959.309" Participant="me006">
	for A_N_N only.
	</Segment>
	<Segment StartTime="1960.665" EndTime="1967.260" Participant="me006">
	And, yeah, the training for the H_T_K models is always, uh, fixed for whatever language you're testing on.
	</Segment>
	<Segment StartTime="1967.314" EndTime="1967.776" Participant="me013">
	Right.
	</Segment>
	<Segment StartTime="1967.776" EndTime="1967.915" Participant="me013">
	<NonVocalSound Description="clunk noise on table"/>
	</Segment>
	<Segment StartTime="1968.791" EndTime="1971.243" Participant="me006">
	And then, there's the testing corpus.
	</Segment>
	<Segment StartTime="1971.243" EndTime="1977.069" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="1977.069" EndTime="1978.246" Participant="me006">
	<VocalSound Description="swallow"/> <VocalSound Description="mouth"/> <VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="1978.246" EndTime="1985.166" Participant="me006">
	So, then I think it's probably instructive to go and - and - and show you the features that we were talking about.
	</Segment>
	<Segment StartTime="1985.166" EndTime="1990.125" Participant="me006">
	Um, so, let's see. Help me out with - With what?
	</Segment>
	<Segment StartTime="1988.264" EndTime="1990.152" Participant="mn007">
	<Uncertain NumSyllables="1"> @@ </Uncertain> P_L_P. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1990.152" EndTime="1990.926" Participant="mn007">
	P_L_P.
	</Segment>
	<Segment StartTime="1990.832" EndTime="1992.241" Participant="me006">
	P_L_P? O_K.
	</Segment>
	<Segment StartTime="1992.241" EndTime="1994.769" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="1993.708" EndTime="1994.769" Participant="mn007">
	M_S_G. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1994.769" EndTime="1995.540" Participant="me006">
	M_S_G. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1995.540" EndTime="1997.068" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="1997.068" EndTime="1998.751" Participant="mn007">
	Uh, J_RASTA. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="1998.751" EndTime="2002.568" Participant="me006">
	J_RASTA. <NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2002.568" EndTime="2004.866" Participant="mn007">
	And J_RASTA-L_D_A. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="2004.866" EndTime="2006.813" Participant="me006">
	J_RASTA-L_D_A.
	</Segment>
	<Segment StartTime="2006.813" EndTime="2011.768" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2008.478" EndTime="2009.862" Participant="mn007">
	Um,
	</Segment>
	<Segment StartTime="2011.194" EndTime="2012.254" Participant="mn007">
	multi-band. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="2012.765" EndTime="2013.937" Participant="me006">
	Multi-band.
	</Segment>
	<Segment StartTime="2013.937" EndTime="2018.778" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2016.495" EndTime="2021.584" Participant="mn007">
	So there would be multi-band before, um - before <Uncertain> our </Uncertain> network, I mean.
	</Segment>
	<Segment StartTime="2022.022" EndTime="2024.197" Participant="me006">
	Yeah, just the multi-band features, right?
	</Segment>
	<Segment StartTime="2022.455" EndTime="2022.914" Participant="mn007">
	And -
	</Segment>
	<Segment StartTime="2024.197" EndTime="2024.607" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2024.607" EndTime="2024.984" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2028.958" EndTime="2030.092" Participant="me013">
	Uh-huh. Ah. Ah.
	</Segment>
	<Segment StartTime="2029.655" EndTime="2032.834" Participant="mn007">
	So, something like, uh, s- <Uncertain> T_C_T </Uncertain> within bands and - Well.
	</Segment>
	<Segment StartTime="2033.860" EndTime="2043.266" Participant="mn007">
	And then multi-band after networks. Meaning that we would have, uh, neural networks, uh, discriminant neural networks for each band.
	</Segment>
	<Segment StartTime="2043.266" EndTime="2044.246" Participant="mn007">
	Uh, yeah. <Comment Description="quietly"/>
	</Segment>
	<Segment StartTime="2044.246" EndTime="2049.905" Participant="mn007">
	And using the - the outputs of these networks or the linear outputs or something like that.
	</Segment>
	<Segment StartTime="2051.201" EndTime="2052.296" Participant="mn007">
	Uh, yeah.
	</Segment>
	<Segment StartTime="2053.007" EndTime="2054.351" Participant="me018">
	What about mel cepstrum?
	</Segment>
	<Segment StartTime="2055.798" EndTime="2057.303" Participant="me006">
	Oh, um -
	</Segment>
	<Segment StartTime="2056.099" EndTime="2059.516" Participant="me018">
	Or is that - you don't include that because it's part of the base or <Uncertain> something </Uncertain>?
	</Segment>
	<Segment StartTime="2059.984" EndTime="2061.722" Participant="me013">
	Well, y- you do have a baseline
	</Segment>
	<Segment StartTime="2060.436" EndTime="2062.532" Participant="fn002">
	Yeah <Uncertain> databases. </Uncertain> Yeah.
	</Segment>
	<Segment StartTime="2062.544" EndTime="2065.300" Participant="me013">
	system that's m- that's mel cepstra, right? So.
	</Segment>
	<Segment StartTime="2064.946" EndTime="2065.751" Participant="fn002">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2066.222" EndTime="2071.028" Participant="mn007">
	But, uh, well, not for the - the A_N_N. I mean -
	</Segment>
	<Segment StartTime="2071.409" EndTime="2071.920" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="2072.296" EndTime="2074.961" Participant="mn007">
	So, yeah, we could - we could add <Pause/> M_F_C_C also.
	</Segment>
	<Segment StartTime="2073.435" EndTime="2076.726" Participant="me006">
	We could add - <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2073.778" EndTime="2081.981" Participant="me013">
	Probably should. I mean at least - at least conceptually, you know, it doesn't meant you actually have to do it, but conceptually it makes sense as a - as a base line.
	</Segment>
	<Segment StartTime="2075.441" EndTime="2076.217" Participant="mn007">
	<Uncertain> Yeah </Uncertain>.
	</Segment>
	<Segment StartTime="2077.084" EndTime="2078.030" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="2081.981" EndTime="2084.421" Participant="me018">
	It'd be an interesting test just to have -
	</Segment>
	<Segment StartTime="2084.421" EndTime="2091.185" Participant="me018">
	just to do M_F_C_C with the neural net and everything else the same. Compare that with just M_- M_F_C_C without the - the net.
	</Segment>
	<Segment StartTime="2085.497" EndTime="2087.070" Participant="fn002">
	Without the - <Comment Description="completing [me018]'s utterance"/> Yeah.
	</Segment>
	<Segment StartTime="2091.174" EndTime="2091.667" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2091.667" EndTime="2092.475" Participant="fn002">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2094.006" EndTime="2096.281" Participant="me006">
	I think - I think Dan did some of that.
	</Segment>
	<Segment StartTime="2096.797" EndTime="2097.194" Participant="me018">
	Oh.
	</Segment>
	<Segment StartTime="2097.331" EndTime="2106.252" Participant="me006">
	Um, in his previous Aurora experiments. And with the net it's - it's wonderful. Without the net it's just baseline.
	</Segment>
	<Segment StartTime="2103.075" EndTime="2104.206" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2105.259" EndTime="2106.033" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2105.867" EndTime="2113.185" Participant="me013">
	Um, I think O_G_I folks have been doing that, too. D- Because I think that for a bunch of their experiments they used, uh, mel cepstra, actually.
	</Segment>
	<Segment StartTime="2108.722" EndTime="2109.297" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2112.252" EndTime="2112.865" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2114.198" EndTime="2114.895" Participant="me013">
	Um,
	</Segment>
	<Segment StartTime="2115.074" EndTime="2118.184" Participant="me013">
	<Uncertain> of </Uncertain> course that's there and this is here and so on. O_K?
	</Segment>
	<Segment StartTime="2118.380" EndTime="2119.149" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="2119.087" EndTime="2119.998" Participant="me034">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="2119.149" EndTime="2124.014" Participant="me006">
	Um, for the training <Pronounce Pronunciation=""corpits""> corpus </Pronounce> - corpus, um, we have,
	</Segment>
	<Segment StartTime="2124.014" EndTime="2127.598" Participant="me006">
	um, the - the d- <Pause/> digits
	</Segment>
	<Segment StartTime="2127.598" EndTime="2128.189" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2128.189" EndTime="2130.779" Participant="me006">
	from the various languages.
	</Segment>
	<Segment StartTime="2130.779" EndTime="2141.749" Participant="me006">
	Um, English <Pause/> Spanish <Pause/> um, French <Pause/> <Comment Description="while writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2144.678" EndTime="2147.077" Participant="me006">
	What else do we have? <VocalSound Description="breath"/>
	</Segment>
	<Segment StartTime="2145.709" EndTime="2148.057" Participant="mn007">
	And the <Pause/> Finnish.
	</Segment>
	<Segment StartTime="2148.057" EndTime="2149.355" Participant="me006">
	Finnish.
	</Segment>
	<Segment StartTime="2148.282" EndTime="2148.913" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2151.381" EndTime="2156.641" Participant="fn002">
	And Italian. Uh, no, Italian no. Italian no. I- Italian yes. Italian?
	</Segment>
	<Segment StartTime="2151.790" EndTime="2154.809" Participant="me018">
	Where did th- where did that come from? Digits? Oh.
	</Segment>
	<Segment StartTime="2154.641" EndTime="2155.372" Participant="me006">
	Oh.
	</Segment>
	<Segment StartTime="2156.479" EndTime="2157.636" Participant="me006">
	Italian.
	</Segment>
	<Segment StartTime="2157.948" EndTime="2158.727" Participant="me013">
	Italian.
	</Segment>
	<Segment StartTime="2158.974" EndTime="2163.486" Participant="me018">
	Is that - Was that distributed with Aurora, or - ? <Comment Description="leading question"/> Where did that - ?
	</Segment>
	<Segment StartTime="2161.979" EndTime="2163.371" Participant="me006">
	One L_ or two L_'s?
	</Segment>
	<Segment StartTime="2162.655" EndTime="2164.570" Participant="me013">
	The newer one. Yeah.
	</Segment>
	<Segment StartTime="2163.320" EndTime="2168.282" Participant="mn007">
	So English, uh, Finnish and Italian are Aurora.
	</Segment>
	<Segment StartTime="2168.282" EndTime="2172.993" Participant="mn007">
	And Spanish and French is something that we can use in addition to Aurora.
	</Segment>
	<Segment StartTime="2173.799" EndTime="2174.642" Participant="mn007">
	Uh, well.
	</Segment>
	<Segment StartTime="2180.685" EndTime="2184.399" Participant="me013">
	Yeah, so Carmen brought the Spanish, and Stephane brought the French.
	</Segment>
	<Segment StartTime="2188.975" EndTime="2190.047" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="2191.008" EndTime="2196.389" Participant="me006">
	And, um, <VocalSound Description="mouth"/> <VocalSound Description="inbreath"/> oh yeah, and -
	</Segment>
	<Segment StartTime="2194.536" EndTime="2197.159" Participant="me013">
	Is it French French or Belgian French? There's a -
	</Segment>
	<Segment StartTime="2196.896" EndTime="2201.218" Participant="mn007">
	<VocalSound Description="laugh"/> It's, uh, French French. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2197.152" EndTime="2200.470" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2197.159" EndTime="2201.248" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2197.342" EndTime="2199.324" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2197.731" EndTime="2199.073" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2200.024" EndTime="2201.265" Participant="me006">
	<Emphasis> French </Emphasis> French. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="2200.470" EndTime="2205.014" Participant="fn002">
	Like Mexican Spain and Spain. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2201.721" EndTime="2205.285" Participant="me013">
	Yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2201.932" EndTime="2203.882" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2204.828" EndTime="2206.877" Participant="me034">
	Or Swiss. Swiss-German.
	</Segment>
	<Segment StartTime="2205.014" EndTime="2210.896" Participant="fn002">
	I think that is more important, Mexican Spain. Because more people - Yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2208.112" EndTime="2209.547" Participant="me013">
	Yeah. Yeah, probably so. Yeah.
	</Segment>
	<Segment StartTime="2209.547" EndTime="2216.513" Participant="me013">
	Yeah, Herve always insists that Belgian is - i- is absolutely pure French, has nothing to do with - but he says those - those - those Parisians talk funny.
	</Segment>
	<Segment StartTime="2216.513" EndTime="2219.292" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2216.927" EndTime="2219.222" Participant="mn007">
	Yeah, yeah, yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2219.222" EndTime="2222.201" Participant="mn007">
	<Emphasis> They </Emphasis> have an accent. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2220.104" EndTime="2222.313" Participant="me013">
	Yeah they - they do, yeah.
	</Segment>
	<Segment StartTime="2222.923" EndTime="2224.531" Participant="me013">
	Yeah. <Pause/>
	</Segment>
	<Segment StartTime="2224.531" EndTime="2227.547" Participant="me013">
	But then he likes Belgian fries too, so. <Pause/> O_K.
	</Segment>
	<Segment StartTime="2229.933" EndTime="2231.344" Participant="me006">
	And then we have, uh,
	</Segment>
	<Segment StartTime="2231.875" EndTime="2232.758" Participant="me006">
	um,
	</Segment>
	<Segment StartTime="2233.179" EndTime="2233.922" Participant="me006">
	broader -
	</Segment>
	<Segment StartTime="2234.301" EndTime="2235.622" Participant="me006">
	broader corpus,
	</Segment>
	<Segment StartTime="2236.284" EndTime="2238.759" Participant="me006">
	um, like TIMIT. <VocalSound Description="outbreath"/>
	</Segment>
	<Segment StartTime="2242.286" EndTime="2243.401" Participant="me006">
	TIMIT so far, right?
	</Segment>
	<Segment StartTime="2242.407" EndTime="2243.522" Participant="fn002">
	And Spanish too.
	</Segment>
	<Segment StartTime="2243.401" EndTime="2244.132" Participant="me006">
	Spanish -
	</Segment>
	<Segment StartTime="2244.132" EndTime="2245.831" Participant="me006">
	Oh, Spanish stories?
	</Segment>
	<Segment StartTime="2247.375" EndTime="2249.251" Participant="fn002">
	Albayzin is the name.
	</Segment>
	<Segment StartTime="2248.861" EndTime="2250.080" Participant="me018">
	What about T_I-digits?
	</Segment>
	<Segment StartTime="2251.297" EndTime="2255.627" Participant="me006">
	Um, T_I-digits - uh all these Aurora f- d- data p-
	</Segment>
	<Segment StartTime="2255.609" EndTime="2255.980" Participant="me018">
	Uh-huh.
	</Segment>
	<Segment StartTime="2255.627" EndTime="2259.099" Participant="me006">
	data is from - is derived from T_I-digits.
	</Segment>
	<Segment StartTime="2257.805" EndTime="2259.317" Participant="me018">
	Oh. Oh O_K.
	</Segment>
	<Segment StartTime="2259.099" EndTime="2265.018" Participant="me006">
	Um, basically, they - they corrupted it with, uh, different kinds of noises at different S_N_R levels.
	</Segment>
	<Segment StartTime="2263.495" EndTime="2264.533" Participant="me018">
	Ah.
	</Segment>
	<Segment StartTime="2265.018" EndTime="2265.593" Participant="me018">
	I see.
	</Segment>
	<Segment StartTime="2266.157" EndTime="2266.472" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2269.095" EndTime="2273.413" Participant="me013">
	y- And I think Stephane was saying there's - there's some broader s- material in the French also?
	</Segment>
	<Segment StartTime="2272.702" EndTime="2274.942" Participant="mn007">
	Yeah, we cou- we could use - Yeah.
	</Segment>
	<Segment StartTime="2274.407" EndTime="2274.942" Participant="me006">
	<VocalSound Description="mouth"/> <Pronounce Pronunciation=""mm-kay""> O_K. </Pronounce>
	</Segment>
	<Segment StartTime="2276.333" EndTime="2277.637" Participant="mn007">
	The French data. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="2278.475" EndTime="2280.691" Participant="fn002">
	Spanish stories? No.
	</Segment>
	<Segment StartTime="2280.691" EndTime="2282.501" Participant="me006">
	No. Sp- Not Spanish stories?
	</Segment>
	<Segment StartTime="2281.315" EndTime="2281.978" Participant="fn002">
	No.
	</Segment>
	<Segment StartTime="2282.501" EndTime="2283.731" Participant="me013">
	Spanish -
	</Segment>
	<Segment StartTime="2282.670" EndTime="2283.465" Participant="fn002">
	No.
	</Segment>
	<Segment StartTime="2283.318" EndTime="2286.003" Participant="me006">
	Spanish <Emphasis> something. </Emphasis> <VocalSound Description="laugh"/> O_K.
	</Segment>
	<Segment StartTime="2283.465" EndTime="2284.292" Participant="fn002">
	Albayz-
	</Segment>
	<Segment StartTime="2284.292" EndTime="2284.974" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="2284.755" EndTime="2287.501" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2286.003" EndTime="2287.668" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2287.668" EndTime="2293.865" Participant="me034">
	Did the Aurora people actually corrupt it themselves, or just specify the signal and the signal-to-noise ratio?
	</Segment>
	<Segment StartTime="2293.795" EndTime="2300.650" Participant="me006">
	They - they corrupted it, um, themselves, but they also included the - the noise files for us, right?
	</Segment>
	<Segment StartTime="2295.565" EndTime="2296.032" Participant="me034">
	O_K.
	</Segment>
	<Segment StartTime="2301.293" EndTime="2301.859" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2301.293" EndTime="2304.599" Participant="me006">
	Or - so we can go ahead and corrupt other things.
	</Segment>
	<Segment StartTime="2302.059" EndTime="2302.641" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="2304.783" EndTime="2307.941" Participant="me013">
	I'm just curious, Carmen - I mean, I couldn't tell if you were joking or -
	</Segment>
	<Segment StartTime="2307.941" EndTime="2310.233" Participant="me013">
	i- Is it - is it Mexican Spanish, or is it -
	</Segment>
	<Segment StartTime="2309.662" EndTime="2314.361" Participant="fn002">
	No no no no. No no no no. Spanish from Spain.
	</Segment>
	<Segment StartTime="2310.233" EndTime="2313.861" Participant="me013">
	Oh, no, no. It's - it's Spanish from Spain, Spanish. Yeah, O_K.
	</Segment>
	<Segment StartTime="2310.751" EndTime="2312.418" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2314.877" EndTime="2317.672" Participant="me006">
	<VocalSound Description="laugh"/> From Spain.
	</Segment>
	<Segment StartTime="2315.208" EndTime="2320.693" Participant="fn002">
	<NonVocalSound Description="mike noises"/>
	</Segment>
	<Segment StartTime="2316.366" EndTime="2317.178" Participant="me013">
	Alright.
	</Segment>
	<Segment StartTime="2317.178" EndTime="2323.044" Participant="me013">
	Spanish from Spain. Yeah, we're really covered there now. O_K. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2320.031" EndTime="2323.208" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2320.693" EndTime="2321.580" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2323.044" EndTime="2324.146" Participant="me013">
	And the French from France.
	</Segment>
	<Segment StartTime="2323.208" EndTime="2324.102" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="2323.544" EndTime="2327.190" Participant="mn007">
	Yeah, the - No, the French is f- yeah, from, uh, Paris, also.
	</Segment>
	<Segment StartTime="2324.102" EndTime="2327.007" Participant="me006">
	<NonVocalSound Description="snaps top on or off of dry marker"/>
	</Segment>
	<Segment StartTime="2324.146" EndTime="2331.305" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2327.007" EndTime="2329.052" Participant="me006">
	Oh, from Paris, O_K.
	</Segment>
	<Segment StartTime="2327.190" EndTime="2330.383" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2332.258" EndTime="2332.783" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="2333.940" EndTime="2334.964" Participant="mn007">
	O_K.
	</Segment>
	<Segment StartTime="2334.125" EndTime="2346.105" Participant="me006">
	And TIMIT's from <Pause/> lots of different places. <VocalSound Description="laugh"/> Yeah. <Comment Description="while laughing"/> Yeah.
	</Segment>
	<Segment StartTime="2335.963" EndTime="2341.484" Participant="me013">
	From T_I. From - i- It's from Texas. So may- maybe it's - <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2341.280" EndTime="2342.221" Participant="me034">
	From the deep South. <Comment Description="mock Southern accent"/>
	</Segment>
	<Segment StartTime="2341.484" EndTime="2344.736" Participant="me013">
	So- s- so it's not really from the U_S either. Is that - ? O_K.
	</Segment>
	<Segment StartTime="2346.105" EndTime="2354.999" Participant="me006">
	O_K. And, um, with- within the training corporas um, we're, uh, thinking about, um, training with noise.
	</Segment>
	<Segment StartTime="2354.999" EndTime="2364.430" Participant="me006">
	So, incorporating the same kinds of noises that, um, Aurora is in- incorporating in their, um - in their training corpus.
	</Segment>
	<Segment StartTime="2364.430" EndTime="2370.487" Participant="me006">
	Um, I don't think we- we're given the, uh - the unseen noise conditions, though, right?
	</Segment>
	<Segment StartTime="2370.487" EndTime="2382.697" Participant="me013">
	I think what they were saying was that, um, for this next test there's gonna be some of the cases where they have the same type of noise as you were given before hand and some cases where you're not.
	</Segment>
	<Segment StartTime="2371.429" EndTime="2372.135" Participant="me006">
	Like -
	</Segment>
	<Segment StartTime="2380.886" EndTime="2381.609" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2383.079" EndTime="2383.765" Participant="mn007">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2383.414" EndTime="2384.194" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="2384.194" EndTime="2391.970" Participant="me013">
	So, presumably, that'll be part of the topic of analysis of the - the test results, is how well you do when it's matching noise and how well you do where it's not.
	</Segment>
	<Segment StartTime="2392.690" EndTime="2393.788" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="2392.853" EndTime="2393.730" Participant="me013">
	I think that's right.
	</Segment>
	<Segment StartTime="2393.788" EndTime="2398.518" Participant="me006">
	So, I guess we can't train on - on the - the unseen noise conditions.
	</Segment>
	<Segment StartTime="2398.518" EndTime="2404.565" Participant="me013">
	Well, not if it's not seen, yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2399.052" EndTime="2401.882" Participant="me006">
	Right. If - Not if it's unseen. Yeah.
	</Segment>
	<Segment StartTime="2404.791" EndTime="2405.255" Participant="me013">
	<Uncertain> O_K. </Uncertain>
	</Segment>
	<Segment StartTime="2405.987" EndTime="2416.711" Participant="me013">
	I mean, i- i- i- i- it does seem to me that a lot of times when you train with something that's at least a little bit noisy it can - it can help you out in other kinds of noise even if it's not matching just because
	</Segment>
	<Segment StartTime="2416.711" EndTime="2419.325" Participant="me013">
	there's some more variance that you've built into things.
	</Segment>
	<Segment StartTime="2419.325" EndTime="2424.626" Participant="me013">
	But, but, uh, uh, exactly how well it will work will
	</Segment>
	<Segment StartTime="2419.362" EndTime="2420.206" Participant="mn007">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2424.626" EndTime="2427.663" Participant="me013">
	depend on how near it is to what you had ahead of time. So.
	</Segment>
	<Segment StartTime="2427.518" EndTime="2428.165" Participant="mn007">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2428.625" EndTime="2432.483" Participant="me013">
	O_K, so that's your training corpus, and then your testing corpus - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="2432.483" EndTime="2439.049" Participant="me006">
	Um, the testing corporas are, um, just, um, the same ones as Aurora testing.
	</Segment>
	<Segment StartTime="2439.049" EndTime="2440.773" Participant="me006">
	<NonVocalSound Description="tapping sounds, writing on whiteboard"/>
	</Segment>
	<Segment StartTime="2440.773" EndTime="2447.194" Participant="me006">
	And, that includes, um, the English
	</Segment>
	<Segment StartTime="2449.063" EndTime="2449.802" Participant="fn002">
	Finnish.
	</Segment>
	<Segment StartTime="2449.950" EndTime="2452.355" Participant="me006">
	Spa- um, Italian.
	</Segment>
	<Segment StartTime="2455.876" EndTime="2457.415" Participant="me006">
	Finnish.
	</Segment>
	<Segment StartTime="2460.267" EndTime="2463.529" Participant="me006">
	Uh, we'r- we're gonna get German, right?
	</Segment>
	<Segment StartTime="2463.529" EndTime="2463.717" Participant="me006">
	Ge- <Comment Description="whispered"/>
	</Segment>
	<Segment StartTime="2464.494" EndTime="2468.270" Participant="me013">
	Well, so, yeah, the final test, <Uncertain> on a </Uncertain> guess, is supposed to be German and Danish, right?
	</Segment>
	<Segment StartTime="2465.074" EndTime="2467.521" Participant="me006">
	At the final test will have German.
	</Segment>
	<Segment StartTime="2465.349" EndTime="2466.576" Participant="mn007">
	Uh, yeah.
	</Segment>
	<Segment StartTime="2468.270" EndTime="2468.692" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="2468.692" EndTime="2470.154" Participant="me006">
	<NonVocalSound Description="snaps top on or off of dry marker"/>
	</Segment>
	<Segment StartTime="2469.542" EndTime="2471.829" Participant="mn007">
	The s- yeah, the <Emphasis> Spanish, </Emphasis> perhaps,
	</Segment>
	<Segment StartTime="2471.471" EndTime="2474.519" Participant="me006">
	Spanish. Oh yeah, we can - we can test on s- Spanish.
	</Segment>
	<Segment StartTime="2471.829" EndTime="2473.146" Participant="mn007">
	we will have. Yeah.
	</Segment>
	<Segment StartTime="2474.042" EndTime="2476.175" Participant="mn007">
	But the - the <Emphasis> Aurora </Emphasis> Spanish, I mean.
	</Segment>
	<Segment StartTime="2476.683" EndTime="2478.079" Participant="me006">
	Oh yeah. Mm-hmm.
	</Segment>
	<Segment StartTime="2478.079" EndTime="2481.042" Participant="me013">
	Oh, there's a - there's Spanish testing in the Aurora?
	</Segment>
	<Segment StartTime="2480.411" EndTime="2482.963" Participant="mn007">
	Uh, not yet, but, uh, yeah,
	</Segment>
	<Segment StartTime="2482.963" EndTime="2484.235" Participant="fn002">
	Yeah, it's preparing.
	</Segment>
	<Segment StartTime="2483.361" EndTime="2491.893" Participant="mn007">
	uh, e- pre- they are preparing it, and, well, according to Hynek it will be - we will have this at the end of November, or -
	</Segment>
	<Segment StartTime="2485.915" EndTime="2486.817" Participant="fn002">
	They are preparing.
	</Segment>
	<Segment StartTime="2492.971" EndTime="2494.157" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="2495.480" EndTime="2499.420" Participant="me013">
	O_K, so, uh, something like seven things in each, uh - each column.
	</Segment>
	<Segment StartTime="2496.639" EndTime="2497.335" Participant="mn007">
	Yeah -
	</Segment>
	<Segment StartTime="2499.420" EndTime="2506.680" Participant="me013">
	So that's, uh, three hundred and forty-three, uh, <VocalSound Description="laugh"/> different systems that are going to be developed.
	</Segment>
	<Segment StartTime="2501.606" EndTime="2506.680" Participant="me006">
	<NonVocalSound Description="snaps top on or off of dry marker"/> <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2501.700" EndTime="2505.638" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2506.680" EndTime="2507.806" Participant="me013">
	There's three of you.
	</Segment>
	<Segment StartTime="2507.950" EndTime="2510.321" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2508.271" EndTime="2514.149" Participant="me013">
	Uh, so that's <Comment Description="while laughing"/> hundred and - <VocalSound Description="laugh"/> hundred and fourteen each. <Comment Description="while laughing"/> What? <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2508.695" EndTime="2513.495" Participant="me006">
	Yeah. One hundred each, about. <Comment Description="while laughing"/> <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2508.727" EndTime="2510.157" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2510.659" EndTime="2512.403" Participant="me026">
	What a- what about noise conditions?
	</Segment>
	<Segment StartTime="2510.819" EndTime="2511.756" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2513.703" EndTime="2514.149" Participant="me026">
	w-
	</Segment>
	<Segment StartTime="2514.149" EndTime="2516.673" Participant="me026">
	Don't we need to put in the column for noise conditions?
	</Segment>
	<Segment StartTime="2517.473" EndTime="2520.141" Participant="me013">
	Are you just trying to be difficult? <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2520.140" EndTime="2522.184" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2520.141" EndTime="2521.468" Participant="me026">
	No, I just don't understand. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2520.323" EndTime="2522.865" Participant="mn007">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2522.184" EndTime="2526.461" Participant="me006">
	Well, th- uh, when - when I put these testings on there, I'm assumi-
	</Segment>
	<Segment StartTime="2522.983" EndTime="2526.079" Participant="me013">
	I'm just kidding. Yeah.
	</Segment>
	<Segment StartTime="2526.461" EndTime="2532.751" Participant="me006">
	There- there's three - three tests. Um, type-A_, type-B_, and type-C_. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="2532.751" EndTime="2538.500" Participant="me006">
	And they're all - they're all gonna be test- tested, um, with one training of the H_T_K system.
	</Segment>
	<Segment StartTime="2538.500" EndTime="2550.207" Participant="me006">
	Um, there's a script that tests all three different types of noise conditions. Test-A_ is like a matched noise. Test-B_ is a - is a slightly mismatched. And test-C_ is a, um, mismatched channel.
	</Segment>
	<Segment StartTime="2550.207" EndTime="2550.333" Participant="me006">
	<NonVocalSound Description="snaps top on or off of dry marker"/>
	</Segment>
	<Segment StartTime="2551.176" EndTime="2554.312" Participant="me026">
	And do we do all our <Pause/> training on <Emphasis> clean </Emphasis> data?
	</Segment>
	<Segment StartTime="2555.034" EndTime="2561.589" Participant="me006">
	Um, no, no, we're - we're gonna be, um, training on the noise files that we do have.
	</Segment>
	<Segment StartTime="2555.778" EndTime="2557.624" Participant="fn002">
	Also, we can clean that.
	</Segment>
	<Segment StartTime="2556.412" EndTime="2557.330" Participant="mn007">
	No.
	</Segment>
	<Segment StartTime="2563.816" EndTime="2565.507" Participant="me013">
	So, um -
	</Segment>
	<Segment StartTime="2566.383" EndTime="2571.576" Participant="me013">
	Yeah, so I guess the question is how long does it take to do a - a training? I mean, it's not totally
	</Segment>
	<Segment StartTime="2572.116" EndTime="2579.931" Participant="me013">
	crazy t- I mean, these are - a lot of these are built-in things <Uncertain> and </Uncertain> we know - we have programs that compute P_L_P, we have M_S_G, we have J_RA- <Comment Description="list"/> you know, a lot of these things
	</Segment>
	<Segment StartTime="2579.931" EndTime="2584.067" Participant="me013">
	will just kind of happen, won't take uh a huge amount of development, it's just trying it out.
	</Segment>
	<Segment StartTime="2580.537" EndTime="2581.052" Participant="me034">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="2584.067" EndTime="2586.156" Participant="me013">
	So, we actually can do quite a few experiments.
	</Segment>
	<Segment StartTime="2586.156" EndTime="2591.312" Participant="me013">
	But how - how long does it take, do we think, for one of these
	</Segment>
	<Segment StartTime="2586.483" EndTime="2586.959" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2591.312" EndTime="2593.571" Participant="me013">
	<Pause/> <Comment Description="2.2 second intra-utterance pause"/>
	</Segment>
	<Segment StartTime="2593.571" EndTime="2594.860" Participant="me013">
	trainings?
	</Segment>
	<Segment StartTime="2594.841" EndTime="2597.103" Participant="me006">
	That's a good question. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2596.673" EndTime="2598.570" Participant="me018">
	What about combinations of things?
	</Segment>
	<Segment StartTime="2598.559" EndTime="2605.705" Participant="me013">
	Oh yeah, that's right. I mean, cuz, so, for instance, I think the major advantage of M_S_G - Yeah, good point.
	</Segment>
	<Segment StartTime="2602.227" EndTime="2603.020" Participant="me006">
	Oh!
	</Segment>
	<Segment StartTime="2603.020" EndTime="2604.723" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2604.723" EndTime="2605.148" Participant="me006">
	Och! <Comment Description="nonword, interjection"/>
	</Segment>
	<Segment StartTime="2605.148" EndTime="2606.670" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2605.705" EndTime="2610.134" Participant="me013">
	A major advantage of M_S_G, I see, th- that we've seen in the past is combined with P_L_P.
	</Segment>
	<Segment StartTime="2610.354" EndTime="2610.966" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="2611.681" EndTime="2612.904" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="2613.700" EndTime="2615.310" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2615.310" EndTime="2620.220" Participant="me006">
	Now, this is turning into a four-dimensional cube? <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2620.220" EndTime="2621.631" Participant="me034">
	Or you just add it to the features.
	</Segment>
	<Segment StartTime="2620.626" EndTime="2621.773" Participant="fn002">
	No. <Uncertain> Here. </Uncertain>
	</Segment>
	<Segment StartTime="2620.639" EndTime="2623.705" Participant="me018">
	Well, you just select multiple things on the one dimension.
	</Segment>
	<Segment StartTime="2621.243" EndTime="2621.866" Participant="me006">
	Just -
	</Segment>
	<Segment StartTime="2623.749" EndTime="2624.675" Participant="me006">
	Oh, yeah. O_K.
	</Segment>
	<Segment StartTime="2626.306" EndTime="2630.040" Participant="me013">
	Yeah, so, I mean, <Pause/> you don't wanna, uh -
	</Segment>
	<Segment StartTime="2630.040" EndTime="2633.570" Participant="me013">
	<NonVocalSound Description="page turning"/>
	</Segment>
	<Segment StartTime="2633.570" EndTime="2642.930" Participant="me013">
	Let's see, seven choose two would - <VocalSound Description="laugh"/> - be, uh, twenty-one different combinations. Um.
	</Segment>
	<Segment StartTime="2642.930" EndTime="2644.606" Participant="me034">
	It's not a complete set of combinations, though, right?
	</Segment>
	<Segment StartTime="2642.930" EndTime="2643.520" Participant="me013">
	Probably -
	</Segment>
	<Segment StartTime="2644.606" EndTime="2645.057" Participant="me013">
	What?
	</Segment>
	<Segment StartTime="2645.057" EndTime="2648.040" Participant="me034">
	It's not a complete set of combinations, though, right?
	</Segment>
	<Segment StartTime="2647.085" EndTime="2648.040" Participant="me006">
	No.
	</Segment>
	<Segment StartTime="2647.384" EndTime="2648.110" Participant="me013">
	Yeah, I <Emphasis> hope </Emphasis> not.
	</Segment>
	<Segment StartTime="2648.310" EndTime="2648.910" Participant="me013">
	Yeah, there's -
	</Segment>
	<Segment StartTime="2648.910" EndTime="2650.675" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2649.847" EndTime="2651.473" Participant="me006">
	That would be -
	</Segment>
	<Segment StartTime="2650.675" EndTime="2656.071" Participant="me013">
	Uh, yeah, so P_L_P and M_S_G I think we definitely wanna try cuz we've had a lot of good experience with putting those together.
	</Segment>
	<Segment StartTime="2656.298" EndTime="2656.872" Participant="fn002">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2657.460" EndTime="2658.104" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="2660.630" EndTime="2661.310" Participant="me013">
	<VocalSound Description="breath"/>
	</Segment>
	<Segment StartTime="2661.310" EndTime="2662.090" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="2662.730" EndTime="2670.359" Participant="me018">
	When you do that, you're increasing the size of the inputs to the net. Do you have to reduce the hidden layer, or something?
	</Segment>
	<Segment StartTime="2666.844" EndTime="2670.359" Participant="me013">
	Well, so - I mean, so i- it doesn't increase the number of trainings.
	</Segment>
	<Segment StartTime="2670.890" EndTime="2676.228" Participant="me018">
	No, no, I'm - I'm just wondering about number of parameters in the net. Do you have to worry about keeping that the same, or - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="2671.619" EndTime="2672.049" Participant="me013">
	But -
	</Segment>
	<Segment StartTime="2677.350" EndTime="2678.527" Participant="me013">
	Uh, I don't <Emphasis> think </Emphasis> so.
	</Segment>
	<Segment StartTime="2678.640" EndTime="2680.127" Participant="me034">
	There's a <Emphasis> computation </Emphasis> limit, though, isn't there?
	</Segment>
	<Segment StartTime="2680.900" EndTime="2681.912" Participant="me013">
	Yeah, I mean, it's just more compu-
	</Segment>
	<Segment StartTime="2682.583" EndTime="2683.488" Participant="me013">
	Excuse me?
	</Segment>
	<Segment StartTime="2683.488" EndTime="2688.187" Participant="me034">
	Isn't there like a limit <Pause/> on the computation load, or d- latency, or something like that for Aurora task?
	</Segment>
	<Segment StartTime="2687.344" EndTime="2689.767" Participant="me013">
	Oh yeah, we haven't talked about any of that at all, have we? <Pause/>
	</Segment>
	<Segment StartTime="2689.767" EndTime="2690.321" Participant="me006">
	No.
	</Segment>
	<Segment StartTime="2689.767" EndTime="2691.778" Participant="me013">
	Yeah, so, there's not really a limit.
	</Segment>
	<Segment StartTime="2691.778" EndTime="2696.457" Participant="me013">
	What it is is that there's - there's, uh - it's just penalty, you know?
	</Segment>
	<Segment StartTime="2696.457" EndTime="2703.681" Participant="me013">
	That - that if you're using, uh, a megabyte, then they'll say that's very nice, but, of course, it will never go on a cheap cell phone.
	</Segment>
	<Segment StartTime="2703.860" EndTime="2704.806" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2704.752" EndTime="2705.761" Participant="me034">
	O_K.
	</Segment>
	<Segment StartTime="2704.806" EndTime="2705.819" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="2707.450" EndTime="2710.933" Participant="me013">
	And, u- uh, I think the computation isn't so much of a problem.
	</Segment>
	<Segment StartTime="2711.140" EndTime="2712.341" Participant="me013">
	I think it's more the memory.
	</Segment>
	<Segment StartTime="2713.095" EndTime="2718.527" Participant="me013">
	Uh, and, expensive cell phones, exa- expensive hand-helds, and so forth, are gonna have lots of memory.
	</Segment>
	<Segment StartTime="2719.110" EndTime="2725.503" Participant="me013">
	So it's just that, uh, these people see the - the cheap cell phones as being still the biggest market, so.
	</Segment>
	<Segment StartTime="2725.503" EndTime="2725.980" Participant="me034">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2726.530" EndTime="2727.268" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="2729.410" EndTime="2733.494" Participant="me013">
	But, yeah, I was just realizing that, actually, it doesn't explode out,
	</Segment>
	<Segment StartTime="2733.494" EndTime="2734.292" Participant="me013">
	um -
	</Segment>
	<Segment StartTime="2737.320" EndTime="2738.640" Participant="me013">
	It's not really two to the seventh.
	</Segment>
	<Segment StartTime="2738.640" EndTime="2744.578" Participant="me013">
	But it's - but - but - i- i- it doesn't really explode out the number of trainings cuz these were all trained individually.
	</Segment>
	<Segment StartTime="2746.262" EndTime="2746.684" Participant="me013">
	Right?
	</Segment>
	<Segment StartTime="2746.684" EndTime="2753.726" Participant="me013">
	So, uh, if you have all of these nets trained some place, then, uh, you can
	</Segment>
	<Segment StartTime="2753.726" EndTime="2754.950" Participant="me013">
	combine
	</Segment>
	<Segment StartTime="2755.352" EndTime="2758.912" Participant="me013">
	their outputs and do the K_L transformation and so forth and -
	</Segment>
	<Segment StartTime="2758.698" EndTime="2759.254" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2758.912" EndTime="2760.470" Participant="me013">
	and, uh -
	</Segment>
	<Segment StartTime="2760.470" EndTime="2764.901" Participant="me013">
	So, what it - it blows out is the number of uh testings.
	</Segment>
	<Segment StartTime="2765.948" EndTime="2771.228" Participant="me013">
	And, you know - and the number of times you do that last part. But that last part, I think, is so - has gotta be pretty quick, so.
	</Segment>
	<Segment StartTime="2771.905" EndTime="2772.940" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="2772.940" EndTime="2774.319" Participant="me013">
	Right? I mean, it's just
	</Segment>
	<Segment StartTime="2775.174" EndTime="2777.236" Participant="me013">
	running the data through -
	</Segment>
	<Segment StartTime="2776.229" EndTime="2777.128" Participant="me006">
	Oh.
	</Segment>
	<Segment StartTime="2777.520" EndTime="2779.712" Participant="me013">
	Well, you gotta do the K_L transformation, but -
	</Segment>
	<Segment StartTime="2777.864" EndTime="2781.109" Participant="me018">
	But wh- what about a net that's trained on multiple <Emphasis> languages, </Emphasis> though?
	</Segment>
	<Segment StartTime="2778.146" EndTime="2778.886" Participant="mn007">
	<Uncertain> Eight - </Uncertain> y-
	</Segment>
	<Segment StartTime="2782.160" EndTime="2787.495" Participant="me018">
	Is that just separate nets for each language then combined, or is that actually one net trained on?
	</Segment>
	<Segment StartTime="2784.723" EndTime="2786.473" Participant="fn002">
	Necessary to put <Uncertain> in </Uncertain>. <Comment Description="whispered"/>
	</Segment>
	<Segment StartTime="2787.629" EndTime="2789.270" Participant="mn007">
	Uh, probably one net. Well.
	</Segment>
	<Segment StartTime="2787.730" EndTime="2788.689" Participant="me013">
	Good question.
	</Segment>
	<Segment StartTime="2790.179" EndTime="2790.977" Participant="mn007">
	Uh.
	</Segment>
	<Segment StartTime="2792.353" EndTime="2793.882" Participant="me013">
	One would <Emphasis> think </Emphasis> one net,
	</Segment>
	<Segment StartTime="2793.280" EndTime="2793.882" Participant="mn007">
	So.
	</Segment>
	<Segment StartTime="2794.950" EndTime="2797.329" Participant="me013">
	but we've - I don't think we've <Emphasis> tested </Emphasis> that. Right?
	</Segment>
	<Segment StartTime="2798.810" EndTime="2802.031" Participant="mn007">
	So, in the broader training corpus we can -
	</Segment>
	<Segment StartTime="2802.031" EndTime="2809.928" Participant="mn007">
	we can use, uh, the three, or, a combination of - of two - two languages.
	</Segment>
	<Segment StartTime="2804.393" EndTime="2805.339" Participant="fn002">
	<Uncertain> Database </Uncertain> three.
	</Segment>
	<Segment StartTime="2813.780" EndTime="2814.468" Participant="me018">
	In one net.
	</Segment>
	<Segment StartTime="2814.930" EndTime="2815.207" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2815.207" EndTime="2815.617" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2817.690" EndTime="2824.084" Participant="me013">
	Yeah, so, I guess the first thing is if w- if we know how much a - how long a - a training takes, <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="2824.084" EndTime="2829.047" Participant="me013">
	if we can train up all these - these combinations,
	</Segment>
	<Segment StartTime="2829.047" EndTime="2835.195" Participant="me013">
	uh, then we can start working on testing of them individually, and in combination. Right?
	</Segment>
	<Segment StartTime="2835.128" EndTime="2836.067" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2835.195" EndTime="2842.177" Participant="me013">
	Because the putting them in combination, I think, is not as much computationally as the r- training of the nets in the first place.
	</Segment>
	<Segment StartTime="2842.858" EndTime="2843.269" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2842.858" EndTime="2843.269" Participant="me013">
	Right?
	</Segment>
	<Segment StartTime="2843.810" EndTime="2846.234" Participant="me013">
	So y- you do have to compute the K_L transformation.
	</Segment>
	<Segment StartTime="2847.043" EndTime="2850.057" Participant="me013">
	Uh, which is a little bit, but it's not too much.
	</Segment>
	<Segment StartTime="2850.492" EndTime="2851.732" Participant="mn007">
	It's not too much, no.
	</Segment>
	<Segment StartTime="2851.278" EndTime="2851.928" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="2852.200" EndTime="2852.906" Participant="me013">
	So it's -
	</Segment>
	<Segment StartTime="2853.405" EndTime="2854.130" Participant="mn007">
	But - Yeah.
	</Segment>
	<Segment StartTime="2854.950" EndTime="2861.354" Participant="mn007">
	But there is the testing also, which implies training, uh, the H_T_K models and, well,
	</Segment>
	<Segment StartTime="2859.223" EndTime="2861.736" Participant="fn002">
	The - the model - the H_T_K model.
	</Segment>
	<Segment StartTime="2861.736" EndTime="2862.653" Participant="me013">
	Uh, right.
	</Segment>
	<Segment StartTime="2862.545" EndTime="2866.825" Participant="mn007">
	it's - yeah. But it's - it's - it's not so long. It <Uncertain NumSyllables="2"> @@ </Uncertain> - Yeah.
	</Segment>
	<Segment StartTime="2863.176" EndTime="2866.646" Participant="me013">
	Right. So if you do have lots of combinations, it's -
	</Segment>
	<Segment StartTime="2866.646" EndTime="2869.527" Participant="me013">
	How long does it take for an, uh, H_T_K training?
	</Segment>
	<Segment StartTime="2869.351" EndTime="2871.882" Participant="mn007">
	It's around six hours, I think.
	</Segment>
	<Segment StartTime="2872.770" EndTime="2874.440" Participant="mn007">
	For training and testing, yeah.
	</Segment>
	<Segment StartTime="2872.770" EndTime="2873.956" Participant="fn002">
	It depends on the -
	</Segment>
	<Segment StartTime="2875.200" EndTime="2878.220" Participant="fn002">
	More than six hours. For the Italian, yes.
	</Segment>
	<Segment StartTime="2875.821" EndTime="2876.401" Participant="mn007">
	<Emphasis> More. </Emphasis>
	</Segment>
	<Segment StartTime="2878.910" EndTime="2879.991" Participant="fn002">
	Maybe one day. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2880.026" EndTime="2881.242" Participant="mn007">
	One day?
	</Segment>
	<Segment StartTime="2880.780" EndTime="2882.052" Participant="me013">
	For H_T_K? Really?
	</Segment>
	<Segment StartTime="2880.859" EndTime="2881.214" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="2882.095" EndTime="2882.372" Participant="fn002">
	Well.
	</Segment>
	<Segment StartTime="2882.709" EndTime="2883.389" Participant="me013">
	Running on <Emphasis> what? </Emphasis>
	</Segment>
	<Segment StartTime="2883.670" EndTime="2885.991" Participant="fn002">
	Uh, M_ - M_F_C_C.
	</Segment>
	<Segment StartTime="2885.516" EndTime="2887.752" Participant="me013">
	No, I'm sorry, ru- running on what machine?
	</Segment>
	<Segment StartTime="2887.822" EndTime="2888.819" Participant="fn002">
	Uh, Ravioli.
	</Segment>
	<Segment StartTime="2889.418" EndTime="2893.788" Participant="me013">
	Uh, I don't know what Ravioli is. Is it - is it an Ultra-five, or is it a - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="2894.081" EndTime="2894.459" Participant="fn002">
	<Uncertain> mmm </Uncertain>
	</Segment>
	<Segment StartTime="2894.844" EndTime="2895.545" Participant="fn002">
	Um.
	</Segment>
	<Segment StartTime="2896.020" EndTime="2896.777" Participant="me018">
	I don't know.
	</Segment>
	<Segment StartTime="2896.020" EndTime="2898.341" Participant="me006">
	I don't know. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="2896.603" EndTime="2900.448" Participant="fn002">
	Who is that? I don't know. I don't know. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="2897.340" EndTime="2898.988" Participant="me034">
	<VocalSound Description="laugh"/> I don't know what a <Pronounce> Ravioli </Pronounce> is.
	</Segment>
	<Segment StartTime="2898.341" EndTime="2900.975" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2898.988" EndTime="2900.776" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2900.975" EndTime="2901.760" Participant="me006">
	I don't know. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="2901.760" EndTime="2903.792" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="2902.220" EndTime="2903.913" Participant="me034">
	We can <Emphasis> check </Emphasis> really <Emphasis> quickly, </Emphasis> I guess.
	</Segment>
	<Segment StartTime="2903.055" EndTime="2909.778" Participant="mn007">
	Yeah, I- I think it's- it's- it's not so long because, well, the T_I-digits test data is about, uh
	</Segment>
	<Segment StartTime="2911.025" EndTime="2912.081" Participant="mn007">
	how many hours?
	</Segment>
	<Segment StartTime="2912.678" EndTime="2916.183" Participant="mn007">
	Uh, th- uh, thirty hours of speech, I think, something like that.
	</Segment>
	<Segment StartTime="2913.802" EndTime="2914.765" Participant="me013">
	It's a few hours.
	</Segment>
	<Segment StartTime="2915.370" EndTime="2915.729" Participant="me034">
	Hmm.
	</Segment>
	<Segment StartTime="2915.509" EndTime="2915.994" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="2917.318" EndTime="2918.299" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="2917.756" EndTime="2919.900" Participant="mn007">
	And it p- Well.
	</Segment>
	<Segment StartTime="2920.234" EndTime="2927.818" Participant="me013">
	Right, so, I mean, clearly, there - there's no way we can even <Emphasis> begin </Emphasis> to do an- any significant amount here unless we use multiple machines.
	</Segment>
	<Segment StartTime="2921.831" EndTime="2923.321" Participant="mn007">
	It's six hours.
	</Segment>
	<Segment StartTime="2928.740" EndTime="2935.577" Participant="me013">
	Right? So - so - w- we - I mean there's plenty of machines here and they're n- they're often not in - in a great - great deal of use. So, I mean, I think it's -
	</Segment>
	<Segment StartTime="2935.577" EndTime="2940.213" Participant="me013">
	it's key that - that the - that you look at, uh,
	</Segment>
	<Segment StartTime="2940.213" EndTime="2942.982" Participant="me013">
	you know, what machines are fast, what machines are used a lot - <Comment Description="list"/>
	</Segment>
	<Segment StartTime="2942.982" EndTime="2945.710" Participant="me013">
	Uh, are we still using P_make? Is that - ?
	</Segment>
	<Segment StartTime="2945.710" EndTime="2948.996" Participant="me006">
	Oh, I don't know how w- how we would <Emphasis> P_make </Emphasis> this, though.
	</Segment>
	<Segment StartTime="2949.270" EndTime="2950.710" Participant="me006">
	Um.
	</Segment>
	<Segment StartTime="2951.050" EndTime="2952.002" Participant="me013">
	Well, you have a -
	</Segment>
	<Segment StartTime="2952.657" EndTime="2959.436" Participant="me013">
	I mean, once you get the basic thing set up, you have just all the - uh, a- all these combinations, right?
	</Segment>
	<Segment StartTime="2955.338" EndTime="2955.915" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2959.594" EndTime="2960.140" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="2960.140" EndTime="2960.743" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="2962.558" EndTime="2970.816" Participant="me013">
	It's - it's - let's say it's six hours or eight hours, or something for the training of H_T_K. How long is it for training of - of, uh, the neural net?
	</Segment>
	<Segment StartTime="2970.434" EndTime="2971.172" Participant="me006">
	The neural net?
	</Segment>
	<Segment StartTime="2971.710" EndTime="2972.462" Participant="me006">
	Um.
	</Segment>
	<Segment StartTime="2974.110" EndTime="2975.600" Participant="mn007">
	I would say two days.
	</Segment>
	<Segment StartTime="2975.600" EndTime="2977.196" Participant="me018">
	Depends on the <Pronounce Pronunciation=""corpus-ease""> corpuses, </Pronounce> right?
	</Segment>
	<Segment StartTime="2976.105" EndTime="2977.110" Participant="fn002">
	It depends.
	</Segment>
	<Segment StartTime="2976.610" EndTime="2977.309" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="2977.102" EndTime="2978.309" Participant="me034">
	It s- also depends on the <Emphasis> net. </Emphasis>
	</Segment>
	<Segment StartTime="2977.585" EndTime="2979.955" Participant="fn002">
	Depends on the <Pronounce> corpus. </Pronounce>
	</Segment>
	<Segment StartTime="2977.658" EndTime="2978.532" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2978.309" EndTime="2979.292" Participant="me034">
	How big is the net?
	</Segment>
	<Segment StartTime="2979.955" EndTime="2986.151" Participant="fn002">
	For Albayzin I trained on neural network, uh, was, um, one day also.
	</Segment>
	<Segment StartTime="2986.530" EndTime="2988.204" Participant="me013">
	Uh, but on what machine?
	</Segment>
	<Segment StartTime="2988.469" EndTime="2989.780" Participant="me006">
	On a SPERT board.
	</Segment>
	<Segment StartTime="2988.829" EndTime="2989.866" Participant="fn002">
	Uh.
	</Segment>
	<Segment StartTime="2990.160" EndTime="2994.144" Participant="fn002">
	I - I think <Uncertain> the neural net </Uncertain> SPERT. Yes.
	</Segment>
	<Segment StartTime="2991.274" EndTime="2993.621" Participant="me006">
	Y- you did a - you did it on a SPERT board.
	</Segment>
	<Segment StartTime="2993.527" EndTime="2996.801" Participant="me013">
	O_K, again, we do have a bunch of SPERT boards.
	</Segment>
	<Segment StartTime="2996.801" EndTime="2997.446" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="2998.146" EndTime="3004.276" Participant="me013">
	And I think there - there - there's - I think you folks are probably go- the ones using them right now.
	</Segment>
	<Segment StartTime="3003.809" EndTime="3004.653" Participant="me006">
	<VocalSound Description="breath-laugh"/>
	</Segment>
	<Segment StartTime="3004.010" EndTime="3006.249" Participant="me018">
	Is it faster to do it on the SPERT, or - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="3006.960" EndTime="3008.912" Participant="me013">
	Uh, don't know. Used to be.
	</Segment>
	<Segment StartTime="3008.169" EndTime="3011.772" Participant="me006">
	It's - it's still a little faster on the SPERTs. Yeah, yeah.
	</Segment>
	<Segment StartTime="3010.232" EndTime="3010.608" Participant="me018">
	Is it?
	</Segment>
	<Segment StartTime="3011.840" EndTime="3021.126" Participant="me006">
	Ad- Adam - Adam did some testing. Or either Adam or - or Dan did some testing and they found that the SPERT board's still - still faster. And
	</Segment>
	<Segment StartTime="3020.767" EndTime="3021.162" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3020.979" EndTime="3021.733" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3021.126" EndTime="3026.944" Participant="me006">
	the benefits is that, you know, you run out of SPERT and then you can do other things on your - your computer, and you don't -
	</Segment>
	<Segment StartTime="3025.445" EndTime="3026.237" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3026.066" EndTime="3026.600" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="3027.051" EndTime="3031.097" Participant="me013">
	So you could be - we have quite a few SPERT boards. You could set up, uh,
	</Segment>
	<Segment StartTime="3031.097" EndTime="3035.765" Participant="me013">
	you know, ten different jobs, or something, to run on SPERT - different SPERT boards
	</Segment>
	<Segment StartTime="3035.765" EndTime="3039.799" Participant="me013">
	and - and have ten other jobs running on different computers.
	</Segment>
	<Segment StartTime="3039.799" EndTime="3044.278" Participant="me013">
	So, it's got to take that sort of thing, or - or we're not going to get through any significant number of these.
	</Segment>
	<Segment StartTime="3041.682" EndTime="3042.567" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="3045.093" EndTime="3045.913" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="3047.630" EndTime="3051.959" Participant="me013">
	So this is - Yeah, I mean, I kind of like this because
	</Segment>
	<Segment StartTime="3051.510" EndTime="3053.480" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3051.959" EndTime="3052.578" Participant="me013">
	what it -
	</Segment>
	<Segment StartTime="3052.578" EndTime="3057.715" Participant="me013">
	No - uh, no, what I <Emphasis> like </Emphasis> about it is we - we - we do have a problem that we have very limited time.
	</Segment>
	<Segment StartTime="3053.480" EndTime="3054.166" Participant="me006">
	<Uncertain> O_K. </Uncertain>
	</Segment>
	<Segment StartTime="3058.340" EndTime="3061.729" Participant="me013">
	You know, so, with very limited time, we actually have
	</Segment>
	<Segment StartTime="3061.729" EndTime="3069.630" Participant="me013">
	really quite a - quite a bit of computational resource available if you, you know, get a look across the institute and how little things are being used.
	</Segment>
	<Segment StartTime="3070.530" EndTime="3071.593" Participant="me013">
	And uh,
	</Segment>
	<Segment StartTime="3071.593" EndTime="3073.787" Participant="me013">
	on the other hand, almost anything
	</Segment>
	<Segment StartTime="3073.787" EndTime="3076.780" Participant="me013">
	that really i- you know, is - is new,
	</Segment>
	<Segment StartTime="3077.350" EndTime="3078.481" Participant="me013">
	where we're saying,
	</Segment>
	<Segment StartTime="3077.475" EndTime="3078.519" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="3078.481" EndTime="3083.128" Participant="me013">
	"Well, let's look at, like we were talking before about, uh, uh,
	</Segment>
	<Segment StartTime="3083.128" EndTime="3086.174" Participant="me013">
	voiced-unvoiced-silence detection features and all those sort -"
	</Segment>
	<Segment StartTime="3086.174" EndTime="3088.404" Participant="me013">
	that's - I think it's a great thing to <Emphasis> go </Emphasis> to.
	</Segment>
	<Segment StartTime="3088.404" EndTime="3094.692" Participant="me013">
	But if it's <Emphasis> new, </Emphasis> then we have this development and - and - and learning process t- to - to go through
	</Segment>
	<Segment StartTime="3094.692" EndTime="3098.055" Participant="me013">
	on <Emphasis> top </Emphasis> of - just the - the - all the - all the work.
	</Segment>
	<Segment StartTime="3098.055" EndTime="3099.878" Participant="me013">
	So, I - I - I don't see how we'd <Emphasis> do </Emphasis> it.
	</Segment>
	<Segment StartTime="3099.878" EndTime="3103.545" Participant="me013">
	So what I like about this is you basically have listed all the things that we already know how to do.
	</Segment>
	<Segment StartTime="3103.545" EndTime="3103.960" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3103.960" EndTime="3108.273" Participant="me013">
	And - and all the kinds of data that we, at this point, already have. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3104.409" EndTime="3104.904" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="3108.890" EndTime="3114.940" Participant="me013">
	And, uh, you're just saying let's look at the outer product of all of these things and see if we can calculate <Pronounce Pronunciation=""em""> them. </Pronounce>
	</Segment>
	<Segment StartTime="3113.573" EndTime="3115.808" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3114.940" EndTime="3120.701" Participant="me013">
	a- a- Am I - am I interpreting this correctly? Is this sort of what - what you're thinking of doing in the short term?
	</Segment>
	<Segment StartTime="3120.430" EndTime="3121.025" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="3121.490" EndTime="3122.266" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="3121.603" EndTime="3122.100" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="3122.266" EndTime="3130.800" Participant="me013">
	So - so then I think it's just the - the missing piece is that you need to, uh, you know - you know, talk to - talk to, uh, Chuck, talk to, uh, Adam,
	</Segment>
	<Segment StartTime="3131.390" EndTime="3135.612" Participant="me013">
	uh, sort out about, uh, what's the best way to really,
	</Segment>
	<Segment StartTime="3135.612" EndTime="3140.002" Participant="me013">
	you know, attack this as a - as a - as a mass problem in terms of using many machines.
	</Segment>
	<Segment StartTime="3140.526" EndTime="3145.917" Participant="me013">
	Uh, and uh, then, you know, set it up in terms of scripts and so forth, and -
	</Segment>
	<Segment StartTime="3145.917" EndTime="3151.265" Participant="me013">
	uh, in - in kind o- some kind of structured way. Uh.
	</Segment>
	<Segment StartTime="3152.280" EndTime="3157.342" Participant="me013">
	Um, and, you know, when we go to, uh, O_G_I next week,
	</Segment>
	<Segment StartTime="3157.342" EndTime="3161.644" Participant="me013">
	uh, we can then present to them, you know, what it is that we're doing.
	</Segment>
	<Segment StartTime="3161.644" EndTime="3170.149" Participant="me013">
	And, uh, we can pull things out of this list that we think they are doing sufficiently, that, you know, we're not - we won't be contributing that much.
	</Segment>
	<Segment StartTime="3166.825" EndTime="3167.658" Participant="me006">
	Mmm.
	</Segment>
	<Segment StartTime="3168.091" EndTime="3168.874" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3170.910" EndTime="3171.404" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="3172.290" EndTime="3173.233" Participant="me013">
	And, uh -
	</Segment>
	<Segment StartTime="3175.290" EndTime="3177.612" Participant="me013">
	Then, uh, like, we're <Emphasis> there. </Emphasis>
	</Segment>
	<Segment StartTime="3178.518" EndTime="3178.987" Participant="me013">
	Yeah?
	</Segment>
	<Segment StartTime="3181.170" EndTime="3182.327" Participant="me034">
	How big are the nets you're using?
	</Segment>
	<Segment StartTime="3183.340" EndTime="3185.465" Participant="me006">
	Um, for the -
	</Segment>
	<Segment StartTime="3185.465" EndTime="3187.844" Participant="me006">
	for nets trained on digits, <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="3187.844" EndTime="3193.116" Participant="me006">
	um, we have been using, uh, four hundred order hidden units. <Comment Description=""on the order of four hundred""/>
	</Segment>
	<Segment StartTime="3193.116" EndTime="3204.796" Participant="me006">
	And, um, for the broader class nets we're - we're going to increase that because the, um, the digits nets only correspond to about twenty phonemes.
	</Segment>
	<Segment StartTime="3204.519" EndTime="3204.962" Participant="me034">
	Uh-huh.
	</Segment>
	<Segment StartTime="3205.410" EndTime="3205.961" Participant="me006">
	So.
	</Segment>
	<Segment StartTime="3207.171" EndTime="3208.102" Participant="me013">
	Broader class?
	</Segment>
	<Segment StartTime="3208.102" EndTime="3212.365" Participant="me006">
	Um, the broader - broader training corpus nets like TIMIT.
	</Segment>
	<Segment StartTime="3212.365" EndTime="3214.007" Participant="me006">
	Um, w- we're gonna -
	</Segment>
	<Segment StartTime="3212.773" EndTime="3216.357" Participant="me013">
	Oh, it's not actually broader class, it's actually finer class, but you mean -
	</Segment>
	<Segment StartTime="3216.357" EndTime="3217.246" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="3217.072" EndTime="3219.038" Participant="me013">
	y- You mean <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3217.246" EndTime="3217.803" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="3217.679" EndTime="3219.406" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3218.405" EndTime="3218.913" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="3219.038" EndTime="3221.163" Participant="me013">
	more classes. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3219.560" EndTime="3224.093" Participant="me006">
	<Emphasis> More </Emphasis> classes. Right, right. More classes. That's what I mean. Mm-hmm.
	</Segment>
	<Segment StartTime="3221.163" EndTime="3223.828" Participant="me013">
	Yeah. Yeah. Yeah.
	</Segment>
	<Segment StartTime="3224.248" EndTime="3225.092" CloseMic="false">
	<NonVocalSound Description="thump"/>
	</Segment>
	<Segment StartTime="3226.210" EndTime="3227.543" Participant="me006">
	And.
	</Segment>
	<Segment StartTime="3228.139" EndTime="3228.745" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="3234.130" EndTime="3237.916" Participant="me013">
	Carmen, did you - do you have something else to add? We - you haven't talked too much, and -
	</Segment>
	<Segment StartTime="3237.916" EndTime="3242.836" Participant="fn002">
	<VocalSound Description="laugh"/> D- I begin to work with the Italian database
	</Segment>
	<Segment StartTime="3242.836" EndTime="3249.837" Participant="fn002">
	to - nnn, to - with the f- front-end and with the H_T_K program
	</Segment>
	<Segment StartTime="3249.837" EndTime="3251.660" Participant="fn002">
	and the <Uncertain NumSyllables="4"> @@ </Uncertain>.
	</Segment>
	<Segment StartTime="3251.660" EndTime="3262.694" Participant="fn002">
	And I trained eh, with the Spanish two neural network with P_L_P and with LogRASTA P_L_P.
	</Segment>
	<Segment StartTime="3262.694" EndTime="3268.331" Participant="fn002">
	I don't know exactly what is better if - if LogRASTA or J_RASTA.
	</Segment>
	<Segment StartTime="3269.540" EndTime="3270.838" Participant="me013">
	Well, um,
	</Segment>
	<Segment StartTime="3271.839" EndTime="3273.092" Participant="me013">
	J_RASTA
	</Segment>
	<Segment StartTime="3274.020" EndTime="3277.681" Participant="me013">
	has the <Emphasis> potential </Emphasis> to do better, but it doesn't always.
	</Segment>
	<Segment StartTime="3278.112" EndTime="3280.601" Participant="me013">
	It's - i- i- J_RASTA is more complicated.
	</Segment>
	<Segment StartTime="3280.601" EndTime="3281.028" Participant="me013">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="3281.028" EndTime="3284.467" Participant="me013">
	It's - it's, uh - instead of doing RASTA with a log,
	</Segment>
	<Segment StartTime="3284.467" EndTime="3287.686" Participant="me013">
	you're doing RASTA with a log-like function
	</Segment>
	<Segment StartTime="3287.686" EndTime="3291.365" Participant="me013">
	that varies depending on a J_ parameter,
	</Segment>
	<Segment StartTime="3291.365" EndTime="3294.564" Participant="me013">
	uh, which is supposed to be sensitive to the amount of noise there is.
	</Segment>
	<Segment StartTime="3294.850" EndTime="3300.711" Participant="me013">
	So, it's sort of like the right transformation to do the filtering in, is dependent on how much <Emphasis> noise </Emphasis> there is.
	</Segment>
	<Segment StartTime="3298.194" EndTime="3298.713" Participant="fn002">
	Hm-hmm.
	</Segment>
	<Segment StartTime="3301.075" EndTime="3302.950" Participant="me013">
	And so in J_RASTA you attempt to <Emphasis> do </Emphasis> that.
	</Segment>
	<Segment StartTime="3303.350" EndTime="3312.664" Participant="me013">
	It's a little complicated because once you do that, you end up in some funny domain and you end up having to do a transformation afterwards, which requires some tables.
	</Segment>
	<Segment StartTime="3312.664" EndTime="3325.379" Participant="me013">
	And, uh, so it's - it's - it's a little messier, uh, there's more ways that it can go wrong, uh, but if - if - if you're careful with it, it can do better. So, it's - So.
	</Segment>
	<Segment StartTime="3314.208" EndTime="3315.258" Participant="fn002">
	Hm-hmm.
	</Segment>
	<Segment StartTime="3323.639" EndTime="3325.379" Participant="fn002">
	<Uncertain> It's a bit - I'll </Uncertain> do better.
	</Segment>
	<Segment StartTime="3326.930" EndTime="3332.291" Participant="fn002">
	Um, and I think to - to -
	</Segment>
	<Segment StartTime="3333.722" EndTime="3341.125" Participant="fn002">
	to recognize the Italian digits with the neural netw- Spanish neural network,
	</Segment>
	<Segment StartTime="3341.125" EndTime="3349.110" Participant="fn002">
	and also to train <Uncertain> another </Uncertain> neural network with the Spanish digits, the database of Spanish digits.
	</Segment>
	<Segment StartTime="3349.110" EndTime="3351.655" Participant="fn002">
	And I working that. <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="3350.844" EndTime="3351.320" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="3352.410" EndTime="3357.296" Participant="fn002">
	But prepa- to prepare the - the database are difficult.
	</Segment>
	<Segment StartTime="3357.296" EndTime="3359.313" Participant="fn002">
	Was for <Emphasis> me, </Emphasis>
	</Segment>
	<Segment StartTime="3359.313" EndTime="3362.251" Participant="fn002">
	n- it was a difficult work last week
	</Segment>
	<Segment StartTime="3362.251" EndTime="3373.120" Participant="fn002">
	with the <Pronounce Pronunciation=""labbels""> labels </Pronounce> because the - the program with the <Pronounce Pronunciation=""labbel""> label </Pronounce> obtained that I have, the Albayzin, is different w- to the <Pronounce Pronunciation=""labbel""> label </Pronounce> to train the neural network.
	</Segment>
	<Segment StartTime="3373.120" EndTime="3379.760" Participant="fn002">
	And <Pause/> <VocalSound Description="mouth"/> that is another work that we must to do, to - to <Uncertain> change </Uncertain>.
	</Segment>
	<Segment StartTime="3380.320" EndTime="3381.460" Participant="me013">
	I - I didn't understand.
	</Segment>
	<Segment StartTime="3381.460" EndTime="3387.737" Participant="fn002">
	Uh, for example Albayzin database was <Pronounce Pronunciation=""labbelled""> labeled </Pronounce> automatically with H_T_K.
	</Segment>
	<Segment StartTime="3388.690" EndTime="3396.609" Participant="fn002">
	It's not hand - it's not <Pronounce Pronunciation=""labbels""> labels </Pronounce> by hand. <Emphasis> Labels. </Emphasis> <Comment Description="corrects pronunciation"/> I'm sorry, I'm sorry. The <Emphasis> labels. </Emphasis> I'm sorry.
	</Segment>
	<Segment StartTime="3390.470" EndTime="3396.636" Participant="me013">
	Oh, "l- <Emphasis> labeled </Emphasis>". I'm sorry, I have a p- I had a problem with <VocalSound Description="laugh"/> the pronunciation. <VocalSound Description="laugh-breath"/>
	</Segment>
	<Segment StartTime="3396.609" EndTime="3397.993" Participant="fn002">
	The labels.
	</Segment>
	<Segment StartTime="3396.636" EndTime="3403.396" Participant="me013">
	Yeah, O_K. So, O_K, so let's start over. So, TI- TIMI- TIMIT's hand-labeled, and - and you're saying about the Spanish?
	</Segment>
	<Segment StartTime="3397.993" EndTime="3399.565" Participant="fn002">
	Oh, also that - Yes.
	</Segment>
	<Segment StartTime="3398.478" EndTime="3399.930" Participant="me034">
	<VocalSound Description="cough"/>
	</Segment>
	<Segment StartTime="3403.256" EndTime="3414.510" Participant="fn002">
	The Spanish labels? That was in different format, that the format for the em - the program to train the neural network.
	</Segment>
	<Segment StartTime="3407.399" EndTime="3408.741" Participant="me013">
	Oh, I see.
	</Segment>
	<Segment StartTime="3414.510" EndTime="3416.764" Participant="fn002">
	I necessary to convert.
	</Segment>
	<Segment StartTime="3416.764" EndTime="3418.438" Participant="fn002">
	And someti- well -
	</Segment>
	<Segment StartTime="3419.670" EndTime="3421.082" Participant="fn002">
	It's - it's - Yeah.
	</Segment>
	<Segment StartTime="3419.670" EndTime="3421.694" Participant="me018">
	So you're just having a problem converting the labels.
	</Segment>
	<Segment StartTime="3421.694" EndTime="3436.966" Participant="fn002">
	Yeah, but n- yes, because they have one program, Feacalc, but no, l- LabeCut, but don't - doesn't, eh, include the H_T_K format to convert.
	</Segment>
	<Segment StartTime="3436.966" EndTime="3437.702" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3437.702" EndTime="3438.411" Participant="fn002">
	And,
	</Segment>
	<Segment StartTime="3438.088" EndTime="3438.411" Participant="me034">
	Hmm.
	</Segment>
	<Segment StartTime="3439.680" EndTime="3457.817" Participant="fn002">
	I don't know what. I ask - e- even I ask to Dan Ellis what I can do that, and h- they - he say me that h- he does- doesn't any - any s- any form to - to do that.
	</Segment>
	<Segment StartTime="3457.817" EndTime="3470.629" Participant="fn002">
	And at the end, I think that with LabeCut I can transfer to <Pronounce Pronunciation=""aksee""> ASCII </Pronounce> format, and H_T_K is an <Pronounce Pronunciation=""aksee""> ASCII </Pronounce> format.
	</Segment>
	<Segment StartTime="3470.629" EndTime="3475.194" Participant="fn002">
	And I m- do another, uh, one program to put
	</Segment>
	<Segment StartTime="3475.194" EndTime="3480.580" Participant="fn002">
	<Pronounce Pronunciation=""aksee""> ASCII </Pronounce> format of H_T_K to ase- ay- ac- <Pronounce Pronunciation=""akskee""> ASCII </Pronounce> format to
	</Segment>
	<Segment StartTime="3481.038" EndTime="3481.599" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3481.649" EndTime="3486.199" Participant="fn002">
	<Uncertain> Exceed </Uncertain> and they used LabCut to - <VocalSound Description="laugh"/> to <Uncertain> pass </Uncertain>.
	</Segment>
	<Segment StartTime="3486.199" EndTime="3487.048" Participant="fn002">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3486.199" EndTime="3487.048" Participant="me013">
	O_K, yeah.
	</Segment>
	<Segment StartTime="3487.048" EndTime="3492.510" Participant="fn002">
	<Uncertain> Actually </Uncertain> that was complicated, but well, I know how we can did that - do that.
	</Segment>
	<Segment StartTime="3487.880" EndTime="3488.812" Participant="me013">
	So you-
	</Segment>
	<Segment StartTime="3492.790" EndTime="3499.233" Participant="me013">
	Sure. So it's just usual kind of uh - sometimes say housekeeping, right? To get these - get these things sorted out.
	</Segment>
	<Segment StartTime="3497.895" EndTime="3498.404" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="3499.233" EndTime="3508.431" Participant="me013">
	So it seems like there's - there's some peculiarities of the, uh - of each of these dimensions that are getting sorted out.
	</Segment>
	<Segment StartTime="3508.431" EndTime="3522.498" Participant="me013">
	And then, um, if - if you work on getting the, uh, assembly lines together, and then the - the pieces sort of get ready to go into the assembly line and gradually can start, you know, start turning the crank, more or less.
	</Segment>
	<Segment StartTime="3522.630" EndTime="3524.012" Participant="me013">
	And, uh,
	</Segment>
	<Segment StartTime="3524.012" EndTime="3531.035" Participant="me013">
	uh, we have a lot more computational capability here than they do at O_G_I, so I think that i- if -
	</Segment>
	<Segment StartTime="3531.035" EndTime="3535.694" Participant="me013">
	What's - what's great about this is it sets it up in a very systematic way, so that, uh,
	</Segment>
	<Segment StartTime="3535.694" EndTime="3542.460" Participant="me013">
	once these - all of these, you know, mundane but real problems get sorted out, we can just start turning the crank and - and
	</Segment>
	<Segment StartTime="3537.834" EndTime="3538.908" Participant="fn002">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3542.460" EndTime="3545.792" Participant="me013">
	push all of us through, and then finally figure out what's best.
	</Segment>
	<Segment StartTime="3546.043" EndTime="3546.561" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="3546.880" EndTime="3557.405" Participant="me006">
	Um, I - I was thinking two things. Uh, the first thing was, um - we - we actually had thought of this as sort of like, um - not - not in stages, <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="3557.405" EndTime="3559.873" Participant="me006">
	but more along the - the time axis. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="3559.873" EndTime="3567.203" Participant="me006">
	Just kind of like one stream at a time, je-je-je-je-je <Comment Description="nonword, vocal gesture, to illustrate stream processing"/> check out the results and - and go that way.
	</Segment>
	<Segment StartTime="3560.582" EndTime="3561.139" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3564.664" EndTime="3565.756" Participant="me013">
	Oh, yeah, yeah, sure.
	</Segment>
	<Segment StartTime="3566.350" EndTime="3576.540" Participant="me013">
	No, I'm just saying, I'm just thinking of it like loops, right? And so, y- y- y- if you had three nested loops, that you have a choice for this, a choice for this, and a choice for that, right? And you're going through them all. That - that's what <Emphasis> I </Emphasis> meant.
	</Segment>
	<Segment StartTime="3568.428" EndTime="3568.899" Participant="me006">
	Uh-huh.
	</Segment>
	<Segment StartTime="3573.090" EndTime="3574.903" Participant="me006">
	Yeah. Mm-hmm.
	</Segment>
	<Segment StartTime="3575.521" EndTime="3576.540" Participant="me006">
	Right, right.
	</Segment>
	<Segment StartTime="3576.540" EndTime="3583.245" Participant="me013">
	And, uh, the thing is that once you get a better handle on how much you can realistically do,
	</Segment>
	<Segment StartTime="3583.245" EndTime="3585.467" Participant="me013">
	uh, um, <VocalSound Description="mouth"/>
	</Segment>
	<Segment StartTime="3585.731" EndTime="3594.506" Participant="me013">
	concurrently on different machines, different SPERTs, and so forth, uh, and you see how long it takes on what machine and so forth,
	</Segment>
	<Segment StartTime="3594.506" EndTime="3603.229" Participant="me013">
	you can stand back from it and say, "O_K, if we look at all these combinations we're talking about, and combinations of combinations, and so forth," you'll probably find you can't do it all.
	</Segment>
	<Segment StartTime="3599.503" EndTime="3600.149" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3603.229" EndTime="3603.822" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="3603.822" EndTime="3605.829" Participant="me013">
	O_K, so then at that point,
	</Segment>
	<Segment StartTime="3605.275" EndTime="3607.175" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3605.829" EndTime="3608.629" Participant="me013">
	uh, we should sort out which ones do we throw away.
	</Segment>
	<Segment StartTime="3608.629" EndTime="3612.388" Participant="me013">
	Which of the combinations across - you know, what are the most likely ones, and -
	</Segment>
	<Segment StartTime="3608.707" EndTime="3609.214" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3612.388" EndTime="3617.572" Participant="me013">
	And, uh, I <Emphasis> still </Emphasis> think we could do a <Emphasis> lot </Emphasis> of them. I mean, it wouldn't surprise me if we could do a hundred of them or something.
	</Segment>
	<Segment StartTime="3617.572" EndTime="3623.424" Participant="me013">
	But, probably when you include all the combinations, you're actually talking about a thousand of them or something, and that's probably more than we can do.
	</Segment>
	<Segment StartTime="3621.163" EndTime="3622.991" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3623.424" EndTime="3625.230" Participant="me013">
	Uh, but a hundred is a lot.
	</Segment>
	<Segment StartTime="3626.120" EndTime="3628.400" Participant="me013">
	And - and, uh, um -
	</Segment>
	<Segment StartTime="3630.822" EndTime="3631.383" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="3631.714" EndTime="3632.256" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="3632.256" EndTime="3635.473" Participant="me006">
	Yeah, and the - the <Emphasis> second </Emphasis> thing was about scratch space. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="3635.473" EndTime="3641.303" Participant="me006">
	And I think you sent an email about, um, e- scratch space for - for people
	</Segment>
	<Segment StartTime="3641.303" EndTime="3644.905" Participant="me006">
	to work on. And I know that, uh, Stephane's
	</Segment>
	<Segment StartTime="3644.905" EndTime="3649.886" Participant="me006">
	working from an N_T machine, so his - his home directory exists somewhere else.
	</Segment>
	<Segment StartTime="3647.806" EndTime="3649.646" Participant="me013">
	His - his stuff is somewhere else, yeah.
	</Segment>
	<Segment StartTime="3650.406" EndTime="3653.138" Participant="me013">
	Yeah, I mean, my point I - I want to -
	</Segment>
	<Segment StartTime="3653.138" EndTime="3659.375" Participant="me013">
	Yeah, thanks for bring it back to that. My - th- I want to clarify my point about that - that - that Chuck repeated in his note.
	</Segment>
	<Segment StartTime="3659.375" EndTime="3659.959" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="3661.080" EndTime="3668.178" Participant="me013">
	We're - over the next year or two, we're gonna be upgrading the networks in this place, but right now they're still all te- pretty much all ten megabit lines.
	</Segment>
	<Segment StartTime="3664.976" EndTime="3665.613" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3668.178" EndTime="3671.545" Participant="me013">
	And we <Emphasis> have </Emphasis> reached the - this - the machines are getting faster and faster.
	</Segment>
	<Segment StartTime="3672.040" EndTime="3678.806" Participant="me013">
	So, it actually has reached the point where it's a significant drag on the time for something to move the data from one place to another.
	</Segment>
	<Segment StartTime="3678.887" EndTime="3679.368" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3679.285" EndTime="3685.306" Participant="me013">
	So, you - you don't w- especially in something with repetitive computation where you're going over it multiple times,
	</Segment>
	<Segment StartTime="3685.306" EndTime="3694.700" Participant="me013">
	you do - don't want to have the - the data that you're working on distant from where it's being - where the computation's being done if you can help it.
	</Segment>
	<Segment StartTime="3694.700" EndTime="3695.340" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3695.340" EndTime="3696.030" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="3696.630" EndTime="3708.068" Participant="me013">
	Now, we are getting more disk for the central file server, which, since it's not a computational server, would seem to be a contradiction to what I just said.
	</Segment>
	<Segment StartTime="3708.068" EndTime="3715.867" Participant="me013">
	But the idea is that, uh, suppose you're working with, uh, this big bunch of multi- multilingual databases.
	</Segment>
	<Segment StartTime="3716.649" EndTime="3720.493" Participant="me013">
	Um, you put them all in the central ser- at the cen- central file server.
	</Segment>
	<Segment StartTime="3720.206" EndTime="3720.713" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="3720.920" EndTime="3730.054" Participant="me013">
	Then, when you're working with something and accessing it many times, you copy the piece of it that you're working with over to some place that's close to where the computation is and then do all the work there.
	</Segment>
	<Segment StartTime="3730.054" EndTime="3734.593" Participant="me013">
	And then that way you - you won't have the - the network - you won't be clogging the network for yourself and others.
	</Segment>
	<Segment StartTime="3735.198" EndTime="3736.101" Participant="me013">
	That's the idea.
	</Segment>
	<Segment StartTime="3735.198" EndTime="3735.574" Participant="me006">
	Mmm.
	</Segment>
	<Segment StartTime="3736.710" EndTime="3739.540" Participant="me013">
	So, uh, it's gonna take us -
	</Segment>
	<Segment StartTime="3740.165" EndTime="3745.521" Participant="me013">
	It may be too late for this, uh, p- precise crunch we're in <Emphasis> now, </Emphasis> but, uh,
	</Segment>
	<Segment StartTime="3745.521" EndTime="3754.577" Participant="me013">
	we're, uh - It's gonna take us a couple weeks at least to get the, uh, uh, the amount of disk we're gonna be getting. We're <Emphasis> actually </Emphasis> gonna get, uh,
	</Segment>
	<Segment StartTime="3754.577" EndTime="3760.057" Participant="me013">
	I think four more, uh, thirty-six gigabyte drives
	</Segment>
	<Segment StartTime="3760.057" EndTime="3766.825" Participant="me013">
	and, uh, put them on another - another disk rack. We ran out of space on the disk rack that we had, so we're getting another disk rack and <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3766.825" EndTime="3769.608" Participant="me013">
	four more drives to share between, uh -
	</Segment>
	<Segment StartTime="3769.608" EndTime="3772.960" Participant="me013">
	<Emphasis> primarily </Emphasis> between this project and the Meetings - Meetings Project.
	</Segment>
	<Segment StartTime="3773.920" EndTime="3774.655" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="3775.550" EndTime="3775.730" Participant="me013">
	<VocalSound Description="mouth"/>
	</Segment>
	<Segment StartTime="3777.050" EndTime="3778.492" Participant="me013">
	But, uh,
	</Segment>
	<Segment StartTime="3779.003" EndTime="3784.802" Participant="me013">
	we've put another - I guess there's another eighteen gigabytes that's - that's in there now to help us with the immediate crunch.
	</Segment>
	<Segment StartTime="3785.640" EndTime="3788.750" Participant="me013">
	But, uh, are you saying -
	</Segment>
	<Segment StartTime="3788.750" EndTime="3797.838" Participant="me013">
	So I don't know where <Pause/> you're - Stephane, where you're doing your computations. If - i- so, you're on an N_T machine, so you're using some external machine to -
	</Segment>
	<Segment StartTime="3798.101" EndTime="3799.955" Participant="mn007">
	Yeah, it, uh -
	</Segment>
	<Segment StartTime="3800.780" EndTime="3802.093" Participant="mn007">
	Well, to -
	</Segment>
	<Segment StartTime="3803.390" EndTime="3805.622" Participant="mn007">
	It's Nutmeg and Mustard, I think, the -
	</Segment>
	<Segment StartTime="3806.808" EndTime="3807.859" Participant="me013">
	Do you know these yet?
	</Segment>
	<Segment StartTime="3807.073" EndTime="3808.154" Participant="mn007">
	I don't know what kind.
	</Segment>
	<Segment StartTime="3807.859" EndTime="3808.273" Participant="me018">
	Nuh-uh. <Comment Description="meaning "no""/>
	</Segment>
	<Segment StartTime="3808.190" EndTime="3809.310" Participant="me013">
	Yeah, O_K.
	</Segment>
	<Segment StartTime="3812.004" EndTime="3817.800" Participant="me013">
	Uh, are these - are these, uh, computational servers, or something? I'm - I've been kind of out of it.
	</Segment>
	<Segment StartTime="3812.004" EndTime="3813.074" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3816.666" EndTime="3818.165" Participant="mn007">
	Yeah, I think, yeah.
	</Segment>
	<Segment StartTime="3818.165" EndTime="3819.040" Participant="mn007">
	I think so.
	</Segment>
	<Segment StartTime="3820.773" EndTime="3821.115" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="3821.115" EndTime="3825.784" Participant="me013">
	Unfortunately, these days my idea of running comput- of computa- doing computation is running a spread sheet. So.
	</Segment>
	<Segment StartTime="3825.784" EndTime="3827.810" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3825.931" EndTime="3826.581" Participant="me006">
	<VocalSound Description="perhaps drink and swallow"/>
	</Segment>
	<Segment StartTime="3826.581" EndTime="3828.682" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3827.154" EndTime="3827.737" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="3827.810" EndTime="3831.957" Participant="me013">
	Uh, haven't been - haven't been doing much computing personally, so.
	</Segment>
	<Segment StartTime="3830.407" EndTime="3832.740" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3831.739" EndTime="3832.486" CloseMic="false">
	<NonVocalSound Description="two clunks"/>
	</Segment>
	<Segment StartTime="3832.649" EndTime="3833.570" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="3834.779" EndTime="3841.386" Participant="me013">
	Yeah, so those are computational servers. So I guess the other question is what disk there i- space there is there on the <Pronounce> computational </Pronounce> servers.
	</Segment>
	<Segment StartTime="3840.992" EndTime="3841.480" Participant="me018">
	Right.
	</Segment>
	<Segment StartTime="3842.010" EndTime="3846.213" Participant="me018">
	Yeah, I'm not sure what's available on - is it - you said Nutmeg and what was the other one?
	</Segment>
	<Segment StartTime="3846.690" EndTime="3847.300" Participant="mn007">
	Mustard. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="3847.300" EndTime="3847.820" Participant="me018">
	Mustard.
	</Segment>
	<Segment StartTime="3848.379" EndTime="3848.897" Participant="me018">
	O_K.
	</Segment>
	<Segment StartTime="3850.480" EndTime="3851.054" Participant="me034">
	Huh.
	</Segment>
	<Segment StartTime="3850.480" EndTime="3852.714" Participant="me013">
	Yeah, Well, you're the - you're the disk czar now. So
	</Segment>
	<Segment StartTime="3852.193" EndTime="3854.093" Participant="fn002">
	<VocalSound Description="cough"/>
	</Segment>
	<Segment StartTime="3852.520" EndTime="3854.469" Participant="me018">
	Right, right. <Uncertain> Well </Uncertain>, I'll check on that.
	</Segment>
	<Segment StartTime="3852.714" EndTime="3854.891" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3854.891" EndTime="3855.456" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="3856.511" EndTime="3866.749" Participant="me013">
	Yeah, so basically, uh, Chuck will be the one who will be sorting out what disk needs to be where, and so on, and I'll be the one who says, "O_K, spend the money."
	</Segment>
	<Segment StartTime="3862.515" EndTime="3862.976" Participant="fn002">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="3867.057" EndTime="3868.739" Participant="me013">
	So. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="3868.739" EndTime="3871.420" Participant="me013">
	Which, I mean, n- these days, uh,
	</Segment>
	<Segment StartTime="3871.420" EndTime="3876.108" Participant="me013">
	if you're talking about scratch space, it doesn't increase the, uh, need for backup,
	</Segment>
	<Segment StartTime="3876.108" EndTime="3884.540" Participant="me013">
	and, uh, I think it's not that big a d- and the - the disks themselves are not that expensive. Right now it's -
	</Segment>
	<Segment StartTime="3884.540" EndTime="3896.036" Participant="me018">
	What you can do, when you're on that machine, is, uh, just go to the slash-scratch directory, and do a D_F minus K_, and it'll tell you if there's space available.
	</Segment>
	<Segment StartTime="3893.327" EndTime="3894.197" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="3896.036" EndTime="3897.834" Participant="me018">
	Uh, and if there is then, uh - <Comment Description="abandoned as clear"/>
	</Segment>
	<Segment StartTime="3897.834" EndTime="3902.745" Participant="me013">
	But wasn't it, uh - I think Dave was saying that he preferred that people didn't
	</Segment>
	<Segment StartTime="3902.745" EndTime="3908.414" Participant="me013">
	put stuff in <Emphasis> slash-scratch. </Emphasis> It's more putting in d- s- X_A or X_B or, right?
	</Segment>
	<Segment StartTime="3907.772" EndTime="3911.866" Participant="me018">
	Well, there's different - there, um, there's - Right.
	</Segment>
	<Segment StartTime="3911.866" EndTime="3916.894" Participant="me018">
	So there's the slash-X_-whatever disks, and then there's slash-scratch.
	</Segment>
	<Segment StartTime="3917.432" EndTime="3921.127" Participant="me018">
	And both of those two kinds are not backed up.
	</Segment>
	<Segment StartTime="3921.410" EndTime="3925.675" Participant="me018">
	And if it's called "slash-scratch", it means it's probably an internal disk to the machine.
	</Segment>
	<Segment StartTime="3926.180" EndTime="3926.764" Participant="me018">
	Um.
	</Segment>
	<Segment StartTime="3927.630" EndTime="3929.671" Participant="me018">
	And so that's the kind of thing where, like if -
	</Segment>
	<Segment StartTime="3929.671" EndTime="3930.496" Participant="me018">
	um,
	</Segment>
	<Segment StartTime="3930.496" EndTime="3935.027" Participant="me018">
	O_K, if you don't have an N_T, but you have a - a - a Unix workstation,
	</Segment>
	<Segment StartTime="3935.027" EndTime="3944.616" Participant="me018">
	and they attach an external disk, <Comment Description="rising"/> it'll be called "slash-X_-something" uh, if it's not backed up and it'll be "slash-D_-something" if it <Emphasis> is </Emphasis> backed up.
	</Segment>
	<Segment StartTime="3944.616" EndTime="3948.556" Participant="me018">
	And if it's inside the machine on the desk, it's called "slash-scratch".
	</Segment>
	<Segment StartTime="3948.556" EndTime="3958.140" Participant="me018">
	But the problem is, if you ever get a new machine, they take your machine away. It's easy to unhook the external disks, put them back on the new machine, but then your slash-scratch is gone.
	</Segment>
	<Segment StartTime="3958.140" EndTime="3963.285" Participant="me018">
	So, you don't wanna put anything in slash-scratch that you wanna keep around for a long period of time.
	</Segment>
	<Segment StartTime="3963.285" EndTime="3968.516" Participant="me018">
	But if it's a copy of, say, some data that's on a server,
	</Segment>
	<Segment StartTime="3968.516" EndTime="3971.475" Participant="me018">
	you can put it on slash-scratch because, um,
	</Segment>
	<Segment StartTime="3971.475" EndTime="3977.070" Participant="me018">
	first of all it's not backed up, and second it doesn't matter if that machine disappears and you get a new machine because you just recopy it to slash-scratch.
	</Segment>
	<Segment StartTime="3977.070" EndTime="3988.423" Participant="me018">
	So tha- that's why I was saying you could check slash-scratch on those - on - on, um, Mustard and - and Nutmeg to see if - if there's space that you could use there.
	</Segment>
	<Segment StartTime="3985.526" EndTime="3986.514" Participant="me013">
	I see.
	</Segment>
	<Segment StartTime="3988.423" EndTime="3993.375" Participant="me018">
	You could also use slash-X_-whatever disks on Mustard and Nutmeg.
	</Segment>
	<Segment StartTime="3992.644" EndTime="3993.375" Participant="mn007">
	Yeah, yeah.
	</Segment>
	<Segment StartTime="3993.990" EndTime="3994.690" Participant="me018">
	Um.
	</Segment>
	<Segment StartTime="3996.120" EndTime="4001.335" Participant="me018">
	Yeah, and we do have - I mean, yeah, so - so you - yeah, it's better to have things local
	</Segment>
	<Segment StartTime="4001.335" EndTime="4005.343" Participant="me018">
	if you're gonna run over them lots of times so you don't
	</Segment>
	<Segment StartTime="4005.343" EndTime="4006.558" Participant="me018">
	have to go to the network.
	</Segment>
	<Segment StartTime="4005.505" EndTime="4011.198" Participant="me013">
	Right, so es- so especially if you're - right, if you're - if you're taking some piece of the training corpus,
	</Segment>
	<Segment StartTime="4011.198" EndTime="4017.355" Participant="me013">
	which usually resides in where Chuck is putting it all on the - on the, uh, file server,
	</Segment>
	<Segment StartTime="4017.355" EndTime="4027.703" Participant="me013">
	uh, then, yeah, it's fine if it's not backed up because if it g- g- gets wiped out or something, y- I mean it <Emphasis> is </Emphasis> backed up on the other disk. So, yeah, O_K.
	</Segment>
	<Segment StartTime="4026.499" EndTime="4027.058" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="4028.860" EndTime="4033.132" Participant="me018">
	Yeah, so, <VocalSound Description="clear throat"/> one of the things that I need to - I've <Emphasis> started </Emphasis> looking at -
	</Segment>
	<Segment StartTime="4033.132" EndTime="4035.758" Participant="me018">
	Uh, is this the appropriate time to talk about the disk space stuff?
	</Segment>
	<Segment StartTime="4035.287" EndTime="4035.758" Participant="me013">
	Sure.
	</Segment>
	<Segment StartTime="4036.170" EndTime="4041.005" Participant="me018">
	I've started looking at, um, disk space. Dan - David, um, put a new,
	</Segment>
	<Segment StartTime="4041.005" EndTime="4043.191" Participant="me018">
	um, drive onto Abbott,
	</Segment>
	<Segment StartTime="4043.191" EndTime="4046.696" Participant="me018">
	that's an <Emphasis> X_ </Emphasis> disk, which means it's not backed up. So,
	</Segment>
	<Segment StartTime="4046.696" EndTime="4049.577" Participant="me018">
	um, I've been going through and copying
	</Segment>
	<Segment StartTime="4049.577" EndTime="4055.876" Participant="me018">
	data that is, you know, some kind of <Emphasis> corpus </Emphasis> stuff usually, that - that we've got on a C_D-ROM or something,
	</Segment>
	<Segment StartTime="4055.876" EndTime="4060.630" Participant="me018">
	onto that new disk to free up space <Pause/> on <Emphasis> other </Emphasis> disks.
	</Segment>
	<Segment StartTime="4060.630" EndTime="4065.708" Participant="me018">
	And, um, so far, um, I've copied a couple of Carmen's,
	</Segment>
	<Segment StartTime="4065.708" EndTime="4066.933" Participant="me018">
	um,
	</Segment>
	<Segment StartTime="4066.933" EndTime="4068.231" Participant="me018">
	databases over there.
	</Segment>
	<Segment StartTime="4068.231" EndTime="4072.902" Participant="me018">
	We haven't deleted them off of the slash-D_C disk that they're on right now in Abbott,
	</Segment>
	<Segment StartTime="4072.902" EndTime="4073.450" Participant="me018">
	um,
	</Segment>
	<Segment StartTime="4074.370" EndTime="4075.779" Participant="me018">
	uh, but we -
	</Segment>
	<Segment StartTime="4075.779" EndTime="4077.247" Participant="me018">
	I would like to go through -
	</Segment>
	<Segment StartTime="4077.247" EndTime="4082.826" Participant="me018">
	sit down with you about some of these <Emphasis> other </Emphasis> ones and see if we can move them onto, um,
	</Segment>
	<Segment StartTime="4082.826" EndTime="4084.040" Participant="me018">
	this new disk also.
	</Segment>
	<Segment StartTime="4083.947" EndTime="4085.250" Participant="mn007">
	Yeah, O_K.
	</Segment>
	<Segment StartTime="4084.040" EndTime="4089.116" Participant="me018">
	There's - there's a lot more space there, and it'll free up more space for doing the experiments and things.
	</Segment>
	<Segment StartTime="4089.116" EndTime="4093.160" Participant="me018">
	So, anything that - that you don't need backed up, we can put on this new disk.
	</Segment>
	<Segment StartTime="4093.580" EndTime="4102.280" Participant="me018">
	Um, but if it's experiments and you're creating files and things that you're gonna need, you probably wanna have those on a disk that's backed up, just in case something <Comment Description="while laughing"/>
	</Segment>
	<Segment StartTime="4102.280" EndTime="4103.537" Participant="me018">
	goes wrong. So.
	</Segment>
	<Segment StartTime="4104.040" EndTime="4104.502" Participant="me018">
	Um <Pause/>
	</Segment>
	<Segment StartTime="4105.070" EndTime="4112.002" Participant="me018">
	So far I've - I've copied a couple of things, but I haven't deleted anything off of the old disk to make room yet.
	</Segment>
	<Segment StartTime="4112.480" EndTime="4118.210" Participant="me018">
	Um, and I haven't looked at the - any of the Aurora stuff, except for the Spanish.
	</Segment>
	<Segment StartTime="4118.760" EndTime="4124.190" Participant="me018">
	So I - I guess I'll need to get together with you and see what data we can move onto the new disk.
	</Segment>
	<Segment StartTime="4123.043" EndTime="4123.967" Participant="mn007">
	Yeah, O_K.
	</Segment>
	<Segment StartTime="4127.260" EndTime="4133.461" Participant="me013">
	Um, yeah, I - I just - an- another question occurred to me is - is what were you folks planning to do about normalization?
	</Segment>
	<Segment StartTime="4134.539" EndTime="4135.493" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4136.375" EndTime="4138.492" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="4138.492" EndTime="4142.460" Participant="mn007">
	Well, we were thinking about using this systematically for all the experiments.
	</Segment>
	<Segment StartTime="4143.146" EndTime="4144.049" Participant="mn007">
	Um.
	</Segment>
	<Segment StartTime="4143.977" EndTime="4145.431" Participant="me013">
	This being - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="4144.463" EndTime="4146.060" Participant="mn007">
	<VocalSound Description="mouth"/> So, but -
	</Segment>
	<Segment StartTime="4146.959" EndTime="4148.280" Participant="mn007">
	Uh.
	</Segment>
	<Segment StartTime="4148.280" EndTime="4161.378" Participant="mn007">
	So that this could be another dimension, but we think perhaps we can use the - the best, uh, um, uh, normalization scheme as O_G_I is using, so, with
	</Segment>
	<Segment StartTime="4161.378" EndTime="4163.817" Participant="mn007">
	parameters that they use there, or - and -
	</Segment>
	<Segment StartTime="4163.498" EndTime="4168.965" Participant="me013">
	Yeah, I think that's a good idea. I mean it's i- i- we - we seem to have enough dimensions as it is. So probably if we <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4163.817" EndTime="4166.223" Participant="mn007">
	u- <VocalSound Description="squeak"/> u-
	</Segment>
	<Segment StartTime="4166.982" EndTime="4168.256" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4167.438" EndTime="4169.581" Participant="mn007">
	Yeah, yeah, yeah. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4168.965" EndTime="4172.692" Participant="me013">
	sort of take their - <Emphasis> probably </Emphasis> the <Pronounce> on-line </Pronounce> - line normalization because then it - <Comment Description="speech error, partly corrected"/>
	</Segment>
	<Segment StartTime="4172.692" EndTime="4179.350" Participant="me013">
	it's - if we do anything else, we're gonna end up having to do on-line normalization too, so we may as well just do on-line normalization.
	</Segment>
	<Segment StartTime="4177.296" EndTime="4178.037" Participant="mn007">
	Mm-hmm.
	</Segment>
	<Segment StartTime="4180.140" EndTime="4180.653" Participant="me013">
	So.
	</Segment>
	<Segment StartTime="4180.999" EndTime="4181.391" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="4181.391" EndTime="4184.310" Participant="me013">
	So that it's plausible for the final thing.
	</Segment>
	<Segment StartTime="4184.310" EndTime="4184.687" Participant="me013">
	Good.
	</Segment>
	<Segment StartTime="4185.810" EndTime="4186.415" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="4187.150" EndTime="4189.718" Participant="me013">
	So, I guess, yeah, th- the other topic -
	</Segment>
	<Segment StartTime="4189.718" EndTime="4196.044" Participant="me013">
	I - maybe we're already there, or almost there, is goals for the - for next week's meeting.
	</Segment>
	<Segment StartTime="4196.697" EndTime="4197.507" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4197.507" EndTime="4198.375" Participant="me013">
	i- i-
	</Segment>
	<Segment StartTime="4198.375" EndTime="4199.694" Participant="me013">
	i- it seems to me that
	</Segment>
	<Segment StartTime="4200.422" EndTime="4204.534" Participant="me013">
	we wanna do is flush out what you put on the board here.
	</Segment>
	<Segment StartTime="4204.534" EndTime="4205.360" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4206.471" EndTime="4209.038" Participant="me013">
	You know, maybe, have it be somewhat visual,
	</Segment>
	<Segment StartTime="4209.772" EndTime="4213.686" Participant="me013">
	a little bit. <VocalSound Description="laugh"/> Uh, so w- we can say what we're doing, yeah.
	</Segment>
	<Segment StartTime="4209.958" EndTime="4210.485" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="4211.221" EndTime="4212.753" Participant="me006">
	Like a s- like a slide?
	</Segment>
	<Segment StartTime="4213.302" EndTime="4213.686" Participant="me006">
	O_K.
	</Segment>
	<Segment StartTime="4213.686" EndTime="4219.924" Participant="me013">
	And, um, also, if you have <Pause/> sorted out, um,
	</Segment>
	<Segment StartTime="4219.924" EndTime="4223.980" Participant="me013">
	this information about how long i- <Emphasis> roughly </Emphasis> how long it takes to do on what and,
	</Segment>
	<Segment StartTime="4223.980" EndTime="4231.740" Participant="me013">
	you know, what we can - how many of these trainings, uh, uh, and testings and so forth that we can realistically do,
	</Segment>
	<Segment StartTime="4231.740" EndTime="4239.656" Participant="me013">
	uh, then one of the big goals of going there next week would be to - to actually settle on which of them we're gonna do.
	</Segment>
	<Segment StartTime="4240.682" EndTime="4244.457" Participant="me013">
	And, uh, when we come back we can charge in and do it.
	</Segment>
	<Segment StartTime="4246.263" EndTime="4246.838" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="4248.188" EndTime="4249.846" Participant="me013">
	Anything else that - I-
	</Segment>
	<Segment StartTime="4249.846" EndTime="4251.567" Participant="me013">
	a- a- <Emphasis> Actually </Emphasis> -
	</Segment>
	<Segment StartTime="4251.567" EndTime="4257.105" Participant="me013">
	started out this - this field trip started off with - with, uh, Stephane talking to Hynek, so you may have - you may have had other goals,
	</Segment>
	<Segment StartTime="4257.105" EndTime="4262.634" Participant="me013">
	uh, for going up, and any- anything else you can think of would be - we should think about <Pause/> accomplishing?
	</Segment>
	<Segment StartTime="4262.634" EndTime="4266.063" Participant="me013">
	I mean, I'm just saying this because <Pause/> maybe there's things we need to do in preparation.
	</Segment>
	<Segment StartTime="4269.632" EndTime="4272.480" Participant="mn007">
	<Uncertain> Oh </Uncertain>, I think basically, this is - this is, uh, yeah.
	</Segment>
	<Segment StartTime="4273.483" EndTime="4273.982" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="4276.253" EndTime="4277.131" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="4278.917" EndTime="4279.658" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4280.250" EndTime="4280.547" Participant="me013">
	Alright.
	</Segment>
	<Segment StartTime="4281.006" EndTime="4284.275" Participant="me013">
	And uh - and the other - the - the last topic I had here
	</Segment>
	<Segment StartTime="4284.599" EndTime="4286.275" Participant="me013">
	was, um, uh d-
	</Segment>
	<Segment StartTime="4286.999" EndTime="4289.249" Participant="me013">
	Dave's fine offer to -
	</Segment>
	<Segment StartTime="4289.249" EndTime="4293.233" Participant="me013">
	to, uh, <Emphasis> do </Emphasis> something <Pause/> <VocalSound Description="laugh"/> on <Emphasis> this. </Emphasis> I mean he's doing -
	</Segment>
	<Segment StartTime="4293.233" EndTime="4297.278" Participant="me013">
	<VocalSound Description="laugh"/> - he's working on other things, <VocalSound Description="laugh"/> but to - to do something on <Emphasis> this </Emphasis> project.
	</Segment>
	<Segment StartTime="4294.331" EndTime="4296.327" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4297.278" EndTime="4302.207" Participant="me013">
	So the question is, "Where - where could we, uh, uh, most use Dave's help?"
	</Segment>
	<Segment StartTime="4305.393" EndTime="4305.699" Participant="fn002">
	<VocalSound Description="clears throat"/>
	</Segment>
	<Segment StartTime="4311.093" EndTime="4312.105" Participant="me013">
	<VocalSound Description="breath-laugh"/>
	</Segment>
	<Segment StartTime="4311.221" EndTime="4315.233" Participant="mn007">
	Um, yeah, I was thinking perhaps if, um,
	</Segment>
	<Segment StartTime="4316.654" EndTime="4320.717" Participant="mn007">
	additionally to all these experiments, <Uncertain> which </Uncertain> is not really research,
	</Segment>
	<Segment StartTime="4321.342" EndTime="4326.397" Participant="mn007">
	well I mean it's, uh, running programs and, um,
	</Segment>
	<Segment StartTime="4322.607" EndTime="4323.165" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4324.678" EndTime="4325.893" Participant="me013">
	<VocalSound Description="breath-laugh"/>
	</Segment>
	<Segment StartTime="4326.397" EndTime="4333.724" Participant="mn007">
	<VocalSound Description="mouth"/> trying to have a closer look at the - perhaps the, um,
	</Segment>
	<Segment StartTime="4333.724" EndTime="4337.887" Participant="mn007">
	<VocalSound Description="mouth"/> speech, uh, noise detection or,
	</Segment>
	<Segment StartTime="4337.887" EndTime="4341.834" Participant="mn007">
	uh, voiced-sound-unvoiced-sound detection and -
	</Segment>
	<Segment StartTime="4343.637" EndTime="4347.188" Participant="mn007">
	Which could be important in - i- for noise - noise -
	</Segment>
	<Segment StartTime="4345.311" EndTime="4348.651" Participant="me018">
	I think that would be a - I think that's a big - big deal.
	</Segment>
	<Segment StartTime="4350.070" EndTime="4352.914" Participant="me018">
	Because the - you know, the thing that Sunil was talking about,
	</Segment>
	<Segment StartTime="4352.914" EndTime="4355.069" Participant="me018">
	uh, with the labels,
	</Segment>
	<Segment StartTime="4355.491" EndTime="4356.402" Participant="me018">
	uh,
	</Segment>
	<Segment StartTime="4356.402" EndTime="4358.866" Participant="me018">
	labeling the database when it got to the noisy stuff?
	</Segment>
	<Segment StartTime="4358.866" EndTime="4359.721" Participant="me018">
	The -
	</Segment>
	<Segment StartTime="4360.560" EndTime="4362.818" Participant="me018">
	That - that really throws things off.
	</Segment>
	<Segment StartTime="4362.818" EndTime="4369.890" Participant="me018">
	You know, having the noise all of a sudden, your - your, um, speech detector, I mean the - the, um -
	</Segment>
	<Segment StartTime="4369.890" EndTime="4375.426" Participant="me018">
	What was it? What was happening with <Uncertain> his thing </Uncertain>? He was running through these models very quickly. He was getting lots of, uh,
	</Segment>
	<Segment StartTime="4370.296" EndTime="4371.311" Participant="me013">
	Mmm.
	</Segment>
	<Segment StartTime="4375.426" EndTime="4379.722" Participant="me018">
	uh <Emphasis> insertions, </Emphasis> is what it was, in his recognitions.
	</Segment>
	<Segment StartTime="4375.871" EndTime="4377.138" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="4379.722" EndTime="4383.948" Participant="me013">
	The only problem - I mean, <Emphasis> maybe </Emphasis> that's the right thing - the only problem I have with it is
	</Segment>
	<Segment StartTime="4383.948" EndTime="4386.536" Participant="me013">
	exactly the same reason why
	</Segment>
	<Segment StartTime="4386.536" EndTime="4388.240" Participant="me013">
	you thought it'd be a <Emphasis> good </Emphasis> thing to do.
	</Segment>
	<Segment StartTime="4388.240" EndTime="4389.690" Participant="me013">
	Um, I - I think that -
	</Segment>
	<Segment StartTime="4391.220" EndTime="4399.656" Participant="me013">
	Let's fall back to that. But I think the first responsibility is sort of to figure out if there's something <Pause/> that, uh, an - an additional -
	</Segment>
	<Segment StartTime="4400.716" EndTime="4405.598" Participant="me013">
	<VocalSound Description="laugh"/> Uh, that's a good thing you - remove the mike. Go ahead, good.
	</Segment>
	<Segment StartTime="4405.598" EndTime="4409.911" Participant="me013">
	Uh, <VocalSound Description="laugh"/> uh.
	</Segment>
	<Segment StartTime="4406.411" EndTime="4408.942" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4407.929" EndTime="4408.846" CloseMic="false">
	<NonVocalSound Description="door closes"/>
	</Segment>
	<Segment StartTime="4409.911" EndTime="4414.800" Participant="me013">
	What an additional clever person could help with when we're really in a crunch for time.
	</Segment>
	<Segment StartTime="4415.725" EndTime="4420.053" Participant="me013">
	Right? Cuz Dave's gonna be around for a long time, right? He's - he's gonna be here for <Emphasis> years. </Emphasis>
	</Segment>
	<Segment StartTime="4418.617" EndTime="4419.377" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="4420.458" EndTime="4420.971" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="4420.561" EndTime="4429.411" Participant="me013">
	And so, um, <Pause/> over years, if he's - if he's interested in, you know, voiced-unvoiced-silence, he could do a lot. But if there - if in fact there's something else <Pause/>
	</Segment>
	<Segment StartTime="4421.416" EndTime="4424.901" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4421.553" EndTime="4424.540" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4429.411" EndTime="4434.611" Participant="me013">
	that he could be doing, that would help us when we're - we're sort of uh strapped for time -
	</Segment>
	<Segment StartTime="4434.611" EndTime="4445.513" Participant="me013">
	We have - we - we've, you know, only, <Pause/> uh, another - another month or two <Pause/> to - you know, with the holidays in the middle of it, um, to - to get a lot done. <Comment Description="embedded"/>
	</Segment>
	<Segment StartTime="4445.873" EndTime="4448.774" Participant="me013">
	If we can think of something - some <Emphasis> piece </Emphasis> of this that's going to be -
	</Segment>
	<Segment StartTime="4449.440" EndTime="4454.026" Participant="me013">
	The very fact that it <Emphasis> is </Emphasis> sort of just work, and i- and it's running programs and so forth,
	</Segment>
	<Segment StartTime="4454.026" EndTime="4461.152" Participant="me013">
	is <Emphasis> exactly </Emphasis> <Emphasis> why </Emphasis> <Pause/> it's possible that it - some piece of <Emphasis> could </Emphasis> be handed to someone to do, because it's not - <Comment Description="abandoned as clear"/>
	</Segment>
	<Segment StartTime="4461.152" EndTime="4467.166" Participant="me013">
	Uh, yeah, so that - that's the question. And we don't have to solve it right this s- second, but if we could think of some -
	</Segment>
	<Segment StartTime="4467.166" EndTime="4472.620" Participant="me013">
	some piece that's - that's well defined, that he could help with, he's expressing a will- willingness to do that.
	</Segment>
	<Segment StartTime="4472.990" EndTime="4474.506" Participant="me018">
	What about training up a, um,
	</Segment>
	<Segment StartTime="4472.990" EndTime="4473.571" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4475.451" EndTime="4476.985" Participant="me018">
	a multilingual net?
	</Segment>
	<Segment StartTime="4478.423" EndTime="4487.594" Participant="mn007">
	Yeah. <VocalSound Description="breathe into mike"/>
	</Segment>
	<Segment StartTime="4478.423" EndTime="4487.594" Participant="fn002">
	Yes, maybe to, mmm, put together the - the label - the labels between TIMIT and Spanish or something like that.
	</Segment>
	<Segment StartTime="4487.594" EndTime="4493.470" Participant="mn007">
	Yeah, so defining the superset, and, uh, joining the data and -
	</Segment>
	<Segment StartTime="4489.126" EndTime="4489.771" Participant="fn002">
	Yes.
	</Segment>
	<Segment StartTime="4493.432" EndTime="4493.835" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="4494.575" EndTime="4495.577" Participant="mn007">
	Mmm.
	</Segment>
	<Segment StartTime="4496.330" EndTime="4496.944" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="4498.161" EndTime="4498.975" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="4500.913" EndTime="4501.840" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4503.700" EndTime="4506.076" Participant="me013">
	Yeah, that's something that needs to be done in any event.
	</Segment>
	<Segment StartTime="4506.234" EndTime="4506.777" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="4507.580" EndTime="4510.032" Participant="me013">
	So what we were just saying is that - that, um -
	</Segment>
	<Segment StartTime="4510.032" EndTime="4520.244" Participant="me013">
	I was arguing for, <Pause/> if possible, coming up with something that - that really <Emphasis> was </Emphasis> development and <Emphasis> wasn't </Emphasis> research because we - we're - we have a time crunch.
	</Segment>
	<Segment StartTime="4520.244" EndTime="4528.086" Participant="me013">
	And so, uh, if there's something that would - would save some time that someone else could do on some other piece, then we should think of that first.
	</Segment>
	<Segment StartTime="4528.086" EndTime="4537.847" Participant="me013">
	See the thing with voiced-unvoiced-silence is I really think that - that it's - to do - to do a - a - a - a <Emphasis> poor </Emphasis> job is - is pretty quick, uh, or, you know, a so-so job.
	</Segment>
	<Segment StartTime="4537.847" EndTime="4542.959" Participant="me013">
	You can - you can - you can throw in a couple fea- we know what - what kinds of features help with it. <Pause/>
	</Segment>
	<Segment StartTime="4542.895" EndTime="4543.223" Participant="fn002">
	Hmm.
	</Segment>
	<Segment StartTime="4542.959" EndTime="4544.565" Participant="me013">
	You can throw something in.
	</Segment>
	<Segment StartTime="4544.565" EndTime="4551.753" Participant="me013">
	You can do pretty well. But I remember, in fact, when <Emphasis> you </Emphasis> were working on that, and you worked on for few months, as I recall, and you got to, say ninety-three percent, and
	</Segment>
	<Segment StartTime="4552.268" EndTime="4553.147" Participant="me018">
	Mm-hmm.
	</Segment>
	<Segment StartTime="4552.890" EndTime="4556.637" Participant="me013">
	getting to ninety-four <Pause/> <VocalSound Description="laugh"/> really really hard. <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4555.834" EndTime="4556.502" Participant="me018">
	Another year.
	</Segment>
	<Segment StartTime="4556.502" EndTime="4558.325" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4556.637" EndTime="4559.629" Participant="me006">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4556.637" EndTime="4559.629" Participant="me013">
	Yeah, yeah. So, um -
	</Segment>
	<Segment StartTime="4559.352" EndTime="4559.629" Participant="me034">
	<VocalSound Description="cough"/>
	</Segment>
	<Segment StartTime="4559.629" EndTime="4560.500" Participant="me013">
	And
	</Segment>
	<Segment StartTime="4561.628" EndTime="4564.246" Participant="me013">
	th- th- the other tricky thing is, since we are,
	</Segment>
	<Segment StartTime="4564.890" EndTime="4565.433" Participant="me013">
	uh,
	</Segment>
	<Segment StartTime="4565.433" EndTime="4571.731" Participant="me013">
	even though we're not - we don't have a strict prohibition on memory size, and - and computational complexity,
	</Segment>
	<Segment StartTime="4571.731" EndTime="4575.352" Participant="me013">
	uh, clearly there's some limitation to it. So if we have to -
	</Segment>
	<Segment StartTime="4575.352" EndTime="4580.085" Participant="me013">
	if we say we have to have a pitch detector, say, if we - if we're trying to
	</Segment>
	<Segment StartTime="4580.085" EndTime="4582.584" Participant="me013">
	incorporate pitch information, or at least some kind of
	</Segment>
	<Segment StartTime="4582.584" EndTime="4585.418" Participant="me013">
	harmonic - harmonicity, or something,
	</Segment>
	<Segment StartTime="4585.418" EndTime="4587.526" Participant="me013">
	this is another whole thing,
	</Segment>
	<Segment StartTime="4587.526" EndTime="4588.802" Participant="me013">
	take a while to develop.
	</Segment>
	<Segment StartTime="4588.802" EndTime="4590.529" Participant="me013">
	Anyway, it's a <Emphasis> very </Emphasis> very interesting topic.
	</Segment>
	<Segment StartTime="4590.529" EndTime="4592.129" Participant="me013">
	I mean, one - I think one of the -
	</Segment>
	<Segment StartTime="4592.129" EndTime="4600.019" Participant="me013">
	a <Emphasis> lot </Emphasis> of people would say, and I think Dan would also, uh, that one of the things wrong with current speech recognition is that we - we really do throw away
	</Segment>
	<Segment StartTime="4600.019" EndTime="4601.790" Participant="me013">
	all the harmonicity information.
	</Segment>
	<Segment StartTime="4601.790" EndTime="4603.801" Participant="me013">
	Uh, we try to get spectral envelopes.
	</Segment>
	<Segment StartTime="4603.801" EndTime="4609.950" Participant="me013">
	Reason for doing that is that most of the information about the phonetic identity is in the spectral envelopes are not in the harmonic detail.
	</Segment>
	<Segment StartTime="4609.950" EndTime="4612.196" Participant="me013">
	But the harmonic detail <Emphasis> does </Emphasis> tell you <Emphasis> something. </Emphasis>
	</Segment>
	<Segment StartTime="4612.196" EndTime="4615.711" Participant="me013">
	Like the fact that there <Emphasis> is </Emphasis> harmonic detail is - is real important.
	</Segment>
	<Segment StartTime="4615.711" EndTime="4616.277" Participant="me013">
	<VocalSound Description="inbreath"/> So.
	</Segment>
	<Segment StartTime="4617.090" EndTime="4617.715" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="4618.330" EndTime="4619.308" Participant="me013">
	So, uh.
	</Segment>
	<Segment StartTime="4619.990" EndTime="4623.654" Participant="me013">
	So I think - Yeah. So - wh- that - so the -
	</Segment>
	<Segment StartTime="4623.654" EndTime="4634.691" Participant="me013">
	the other suggestion that just came up was, well what about having him <Pause/> work on the, uh, <Pause/> multilingual super f- superset <Pause/> kind of thing.
	</Segment>
	<Segment StartTime="4634.691" EndTime="4641.885" Participant="me013">
	Uh, coming up with that and then, you know, training it - training a net on that, say, um, from - from, uh - from TIMIT or something.
	</Segment>
	<Segment StartTime="4642.426" EndTime="4644.939" Participant="me013">
	Is that - or uh, for multiple databases. What - what would you -
	</Segment>
	<Segment StartTime="4644.939" EndTime="4646.002" Participant="me013">
	what would you think it would -
	</Segment>
	<Segment StartTime="4646.002" EndTime="4647.992" Participant="me013">
	wh- what would this task consist of?
	</Segment>
	<Segment StartTime="4651.033" EndTime="4663.570" Participant="mn007">
	Yeah, it would consist in, uh, well, um, creating the - the superset, and, uh, modifying the lab- labels for matching the superset.
	</Segment>
	<Segment StartTime="4665.005" EndTime="4666.870" Participant="mn007">
	Uh.
	</Segment>
	<Segment StartTime="4667.940" EndTime="4674.305" Participant="me013">
	Uh, creating a superset from looking at the multiple languages, and then creating i- m- changing labels on <Emphasis> TIMIT? </Emphasis>
	</Segment>
	<Segment StartTime="4668.644" EndTime="4671.087" Participant="mn007">
	Well, creating the mappings, actually. Yeah.
	</Segment>
	<Segment StartTime="4674.305" EndTime="4678.822" Participant="me013">
	Or on - or on multiple language - <VocalSound Description="hiccup"/> multiple languages?
	</Segment>
	<Segment StartTime="4674.989" EndTime="4676.086" Participant="fn002">
	No.
	</Segment>
	<Segment StartTime="4676.292" EndTime="4679.480" Participant="mn007">
	Yeah, yeah, with the <Uncertain NumSyllables="1"> @@ </Uncertain> three languages, and -
	</Segment>
	<Segment StartTime="4676.327" EndTime="4681.770" Participant="fn002">
	The multiple language. Maybe for the other language because TIMIT have more phone.
	</Segment>
	<Segment StartTime="4682.040" EndTime="4682.558" Participant="me013">
	Yeah. <Comment Description="during mike noise"/>
	</Segment>
	<Segment StartTime="4684.675" EndTime="4685.566" Participant="me013">
	Uh.
	</Segment>
	<Segment StartTime="4684.840" EndTime="4688.767" Participant="me018">
	So you'd have to create a mapping from each language to the superset.
	</Segment>
	<Segment StartTime="4686.499" EndTime="4687.904" Participant="fn002">
	Yeah. Mm-hmm.
	</Segment>
	<Segment StartTime="4687.015" EndTime="4689.410" Participant="mn007">
	From each language to the superset, yeah. Mmm.
	</Segment>
	<Segment StartTime="4688.654" EndTime="4689.288" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="4690.110" EndTime="4691.213" Participant="me006">
	There's, um -
	</Segment>
	<Segment StartTime="4690.121" EndTime="4691.117" Participant="me013">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="4691.890" EndTime="4696.814" Participant="me006">
	Carmen was talking about this SAMPA thing, <Comment Description="rising"/> and it's, um,
	</Segment>
	<Segment StartTime="4696.814" EndTime="4702.424" Participant="me006">
	<VocalSound Description="mouth"/> it's an effort by linguists to come up with, um,
	</Segment>
	<Segment StartTime="4702.424" EndTime="4707.580" Participant="me006">
	a machine readable I_P_A, um, sort of thing, right?
	</Segment>
	<Segment StartTime="4707.580" EndTime="4715.104" Participant="me006">
	And, um, they - they have a web site that Stephane was showing us that has, um -
	</Segment>
	<Segment StartTime="4707.921" EndTime="4708.397" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4715.104" EndTime="4729.794" Participant="me006">
	has all the English phonemes and their SAMPA correspondent, um, phoneme, and then, um, they have Spanish, they have German, they have all - all sorts of languages, um, mapping - mapping to the SAMPA phonemes, which -
	</Segment>
	<Segment StartTime="4729.684" EndTime="4742.475" Participant="fn002">
	Yeah, the tr- the transcription, though, for Albayzin is n- the transcription are of SAMPA the same, uh, how you say, symbol that SAMPA appear.
	</Segment>
	<Segment StartTime="4730.636" EndTime="4731.227" Participant="me034">
	SAMPA?
	</Segment>
	<Segment StartTime="4733.355" EndTime="4734.547" Participant="me034">
	What does "SAMPA" mean?
	</Segment>
	<Segment StartTime="4738.517" EndTime="4739.505" Participant="me013">
	Mm-hmm.
	</Segment>
	<Segment StartTime="4740.723" EndTime="4741.483" Participant="me013">
	Hmm.
	</Segment>
	<Segment StartTime="4742.475" EndTime="4746.470" Participant="fn002">
	But I don't know if TIMIT o- how is TIMIT.
	</Segment>
	<Segment StartTime="4746.470" EndTime="4747.056" Participant="me034">
	So, I mean -
	</Segment>
	<Segment StartTime="4746.658" EndTime="4746.994" Participant="me013">
	What -
	</Segment>
	<Segment StartTime="4747.441" EndTime="4747.993" Participant="me034">
	I'm sorry. <Comment Description="during mike noise"/>
	</Segment>
	<Segment StartTime="4747.478" EndTime="4748.194" Participant="me013">
	Go ahead.
	</Segment>
	<Segment StartTime="4749.240" EndTime="4752.511" Participant="me034">
	I was gonna say, does that mean I_P_A is not really international? <VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4752.650" EndTime="4756.134" Participant="me006">
	No, it's - it's saying - y-
	</Segment>
	<Segment StartTime="4753.731" EndTime="4757.655" Participant="me018">
	It uses special diacritics and stuff, which you can't do with ASCII characters.
	</Segment>
	<Segment StartTime="4755.214" EndTime="4755.913" Participant="fn002">
	Yeah.
	</Segment>
	<Segment StartTime="4756.134" EndTime="4757.840" Participant="me006">
	can't print on ASCII.
	</Segment>
	<Segment StartTime="4757.698" EndTime="4758.726" Participant="me034">
	Oh, I see. <Comment Description="during mike noise"/>
	</Segment>
	<Segment StartTime="4758.085" EndTime="4760.584" Participant="me018">
	So the SAMPA's just mapping those.
	</Segment>
	<Segment StartTime="4760.504" EndTime="4760.894" Participant="me034">
	Got it.
	</Segment>
	<Segment StartTime="4762.021" EndTime="4765.970" Participant="me013">
	What, uh - <VocalSound Description="two vocal squeaks"/> Has O_G_I done anything <Emphasis> about </Emphasis> this issue?
	</Segment>
	<Segment StartTime="4765.970" EndTime="4766.704" Participant="me013">
	Do they have -
	</Segment>
	<Segment StartTime="4767.827" EndTime="4770.788" Participant="me013">
	Do they <Emphasis> have </Emphasis> any kind of superset that they already <Emphasis> have? </Emphasis>
	</Segment>
	<Segment StartTime="4771.492" EndTime="4773.313" Participant="mn007">
	I don't think so. Well,
	</Segment>
	<Segment StartTime="4773.905" EndTime="4780.560" Participant="mn007">
	they - they - they're going actually the - the other way, defining uh, phoneme clusters, apparently. Well.
	</Segment>
	<Segment StartTime="4780.508" EndTime="4781.260" Participant="me013">
	Aha.
	</Segment>
	<Segment StartTime="4782.084" EndTime="4782.790" Participant="me013">
	That's right.
	</Segment>
	<Segment StartTime="4784.120" EndTime="4787.818" Participant="me013">
	Uh, and that's an interesting <Pause/> way to go <Emphasis> too. </Emphasis>
	</Segment>
	<Segment StartTime="4786.771" EndTime="4794.287" Participant="me018">
	So they just throw the speech from all different languages together, then cluster it into sixty or fifty or whatever clusters?
	</Segment>
	<Segment StartTime="4790.092" EndTime="4791.860" Participant="mn007">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="4793.364" EndTime="4797.860" Participant="mn007">
	I think they've not done it, uh, doing, uh, multiple language yet, but
	</Segment>
	<Segment StartTime="4798.597" EndTime="4810.661" Participant="mn007">
	what they did is to training, uh, English nets with all the phonemes, and then training <Uncertain> it in </Uncertain> English nets with, uh, kind of seventeen, I think it was - seventeen, uh, broad classes.
	</Segment>
	<Segment StartTime="4809.138" EndTime="4811.010" Participant="me018">
	Automatically derived - Mm-hmm.
	</Segment>
	<Segment StartTime="4812.140" EndTime="4812.788" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="4812.624" EndTime="4814.745" Participant="me018">
	Automatically derived broad classes, or - ? <Comment Description="leading question"/> Uh-huh.
	</Segment>
	<Segment StartTime="4813.649" EndTime="4814.745" Participant="mn007">
	Yeah, I think so.
	</Segment>
	<Segment StartTime="4815.328" EndTime="4817.310" Participant="mn007">
	Uh, and, yeah.
	</Segment>
	<Segment StartTime="4817.659" EndTime="4823.738" Participant="mn007">
	And the result was that apparently, when testing on cross-language it was better. I think so.
	</Segment>
	<Segment StartTime="4824.665" EndTime="4830.328" Participant="mn007">
	But Hynek didn't add - didn't have all the results when he showed me that, so, well. But -
	</Segment>
	<Segment StartTime="4829.772" EndTime="4836.387" Participant="me013">
	So that does make an interesting question, though. Is there's some way that we should tie into that with this. Um.
	</Segment>
	<Segment StartTime="4834.327" EndTime="4836.133" Participant="me034">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="4836.854" EndTime="4837.861" Participant="me034">
	<NonVocalSound Description="mike noise"/>
	</Segment>
	<Segment StartTime="4840.848" EndTime="4849.422" Participant="me013">
	Right? I mean, if - if in fact that is a better thing to do, <Pause/> should we leverage that, rather than doing, <Pause/> um, our own.
	</Segment>
	<Segment StartTime="4851.421" EndTime="4852.020" Participant="me013">
	Right?
	</Segment>
	<Segment StartTime="4852.020" EndTime="4855.455" Participant="me013">
	<VocalSound Description="inbreath"/> So, if i- if - if they s- I mean, we have - <Pause/>
	</Segment>
	<Segment StartTime="4855.455" EndTime="4859.170" Participant="me013">
	i- we have the - the trainings with our own categories.
	</Segment>
	<Segment StartTime="4859.842" EndTime="4862.321" Participant="me013">
	And now we're saying, "Well, how do we handle cross-language?"
	</Segment>
	<Segment StartTime="4862.321" EndTime="4864.827" Participant="me013">
	And one way is to come up with a <Emphasis> superset, </Emphasis> <Pause/>
	</Segment>
	<Segment StartTime="4864.827" EndTime="4868.368" Participant="me013">
	but they are als- they're trying coming up with <Emphasis> clustered, </Emphasis> <Pause/>
	</Segment>
	<Segment StartTime="4868.368" EndTime="4870.827" Participant="me013">
	and do we think there's something <Emphasis> wrong </Emphasis> with that?
	</Segment>
	<Segment StartTime="4870.746" EndTime="4873.552" Participant="mn007">
	I think that there's something wrong or - Well, because -
	</Segment>
	<Segment StartTime="4872.245" EndTime="4873.146" Participant="me013">
	O_K. What w-
	</Segment>
	<Segment StartTime="4873.552" EndTime="4884.788" Participant="mn007">
	Well, for the moment we are testing on digits, and e- i- perhaps u- using broad phoneme classes, it's - it's O_K for um, uh classifying the digits, but
	</Segment>
	<Segment StartTime="4884.788" EndTime="4887.398" Participant="mn007">
	as soon as you will have more words,
	</Segment>
	<Segment StartTime="4887.398" EndTime="4891.555" Participant="mn007">
	well, words can differ with only a single phoneme, and -
	</Segment>
	<Segment StartTime="4891.555" EndTime="4894.352" Participant="mn007">
	which could be the same, uh, class. Well.
	</Segment>
	<Segment StartTime="4893.976" EndTime="4894.590" Participant="me013">
	I see.
	</Segment>
	<Segment StartTime="4895.510" EndTime="4895.805" Participant="mn007">
	So.
	</Segment>
	<Segment StartTime="4898.746" EndTime="4905.113" Participant="me013">
	Right. Although, you are not using this for the - You're using this for the <Emphasis> feature </Emphasis> generation, though, not the -
	</Segment>
	<Segment StartTime="4900.572" EndTime="4901.981" Participant="mn007">
	So, I'm <Uncertain> afraid </Uncertain> -
	</Segment>
	<Segment StartTime="4905.830" EndTime="4911.646" Participant="mn007">
	Yeah, but you will ask the net to put one for th- th- the phoneme class and -
	</Segment>
	<Segment StartTime="4906.444" EndTime="4907.146" Participant="me013">
	s- <Comment Description="whispered"/>
	</Segment>
	<Segment StartTime="4909.399" EndTime="4910.261" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4912.850" EndTime="4913.571" Participant="mn007">
	So.
	</Segment>
	<Segment StartTime="4913.634" EndTime="4914.075" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4913.998" EndTime="4918.120" Participant="me018">
	So you're saying that there may not be enough information coming out of the net to help you discriminate the words?
	</Segment>
	<Segment StartTime="4914.250" EndTime="4915.142" Participant="mn007">
	Well.
	</Segment>
	<Segment StartTime="4916.773" EndTime="4918.120" Participant="mn007">
	Yeah, yeah. Mmm.
	</Segment>
	<Segment StartTime="4919.780" EndTime="4923.140" Participant="me034">
	Fact, most confusions are <Emphasis> within </Emphasis> the phone - phone classes, right?
	</Segment>
	<Segment StartTime="4924.419" EndTime="4928.235" Participant="me034">
	I think, uh, Larry was saying like obstruents are only confused with other obstruents, et cetera, et cetera. <Comment Description="rising"/>
	</Segment>
	<Segment StartTime="4924.570" EndTime="4925.265" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4927.439" EndTime="4928.235" Participant="me018">
	Hmm.
	</Segment>
	<Segment StartTime="4928.235" EndTime="4929.390" Participant="me013">
	Yeah. Yeah.
	</Segment>
	<Segment StartTime="4929.390" EndTime="4931.470" Participant="mn007">
	Yeah, this is another p- yeah, another point.
	</Segment>
	<Segment StartTime="4931.470" EndTime="4932.120" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="4932.693" EndTime="4937.428" Participant="me006">
	So - so, maybe we could look at articulatory type stuff, right?
	</Segment>
	<Segment StartTime="4936.157" EndTime="4939.059" Participant="me013">
	But that's what I thought they were gonna - Did they <Emphasis> not </Emphasis> do that, or - ?
	</Segment>
	<Segment StartTime="4941.466" EndTime="4942.735" Participant="mn007">
	I don't think so. Well,
	</Segment>
	<Segment StartTime="4943.122" EndTime="4946.636" Participant="mn007">
	they were talking about, perhaps, but they d- I d- w- Yeah.
	</Segment>
	<Segment StartTime="4943.157" EndTime="4943.655" Participant="me013">
	So -
	</Segment>
	<Segment StartTime="4944.873" EndTime="4950.939" Participant="me013">
	They're talking about it, but that's sort of a question whether they <Emphasis> did </Emphasis> because that's - that's the <Emphasis> other </Emphasis> route to go. Instead of this, you know -
	</Segment>
	<Segment StartTime="4948.965" EndTime="4949.719" Participant="me006">
	Mm-hmm.
	</Segment>
	<Segment StartTime="4950.658" EndTime="4951.795" Participant="me006">
	Superclass.
	</Segment>
	<Segment StartTime="4951.408" EndTime="4955.267" Participant="me013">
	Instead of the - the - the - the <Emphasis> superclass </Emphasis> thing, which is to take -
	</Segment>
	<Segment StartTime="4956.610" EndTime="4963.620" Participant="me013">
	So suppose y- you don't really mark arti- To <Emphasis> really </Emphasis> mark articulatory features, you really wanna look at the acoustics and - and see where everything is, and we're not gonna do that.
	</Segment>
	<Segment StartTime="4963.620" EndTime="4974.907" Participant="me013">
	So, <Pause/> uh, the second class way of doing it is <Pause/> to look at the, uh, phones that are labeled and translate them into acoustic - uh, uh - articulatory, uh, uh, features.
	</Segment>
	<Segment StartTime="4964.463" EndTime="4966.049" Participant="me034">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="4969.578" EndTime="4970.227" CloseMic="false">
	<NonVocalSound Description="tap"/>
	</Segment>
	<Segment StartTime="4975.520" EndTime="4979.330" Participant="me013">
	So it won't really be right. You won't really have these overlapping <Pause/> things and so forth, but -
	</Segment>
	<Segment StartTime="4978.473" EndTime="4980.613" Participant="me018">
	So the targets of the net - are these - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="4980.748" EndTime="4981.855" Participant="me013">
	<VocalSound Description="mouth"/> Articulatory feature.
	</Segment>
	<Segment StartTime="4981.153" EndTime="4982.476" Participant="me018">
	Articulatory features.
	</Segment>
	<Segment StartTime="4982.004" EndTime="4982.476" Participant="me013">
	Right.
	</Segment>
	<Segment StartTime="4982.476" EndTime="4985.553" Participant="me018">
	But that implies that you can have more than one on at a time?
	</Segment>
	<Segment StartTime="4984.974" EndTime="4985.553" Participant="me013">
	That's right.
	</Segment>
	<Segment StartTime="4985.553" EndTime="4986.346" Participant="me018">
	Ah. O_K.
	</Segment>
	<Segment StartTime="4985.923" EndTime="4987.861" Participant="me013">
	You either do that or you have multiple nets.
	</Segment>
	<Segment StartTime="4987.861" EndTime="4988.322" Participant="me018">
	I see.
	</Segment>
	<Segment StartTime="4989.033" EndTime="4989.784" Participant="me013">
	Um.
	</Segment>
	<Segment StartTime="4990.220" EndTime="4991.515" Participant="me013">
	And, um
	</Segment>
	<Segment StartTime="4992.660" EndTime="4994.553" Participant="me013">
	I don't know if our software -
	</Segment>
	<Segment StartTime="4995.549" EndTime="4998.746" Participant="me013">
	this - if the qu- versions of the Quicknet that we're using allows for that.
	</Segment>
	<Segment StartTime="4998.746" EndTime="4999.377" Participant="me013">
	Do you know?
	</Segment>
	<Segment StartTime="4999.660" EndTime="5001.042" Participant="me006">
	Allows for - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="5000.585" EndTime="5002.887" Participant="me013">
	Multiple targets being one?
	</Segment>
	<Segment StartTime="5002.640" EndTime="5007.821" Participant="me006">
	Oh, um, we have gotten soft targets to - to work.
	</Segment>
	<Segment StartTime="5006.602" EndTime="5008.553" Participant="me013">
	O_K. So that - that'll work, yeah.
	</Segment>
	<Segment StartTime="5008.553" EndTime="5009.500" Participant="me006">
	Yeah.
	</Segment>
	<Segment StartTime="5009.500" EndTime="5009.938" Participant="me013">
	O_K.
	</Segment>
	<Segment StartTime="5010.740" EndTime="5018.486" Participant="me013">
	So, um, that's <Emphasis> another </Emphasis> thing that could be done - is that we could - we could, uh, just translate - instead of translating to a superset, <Pause/>
	</Segment>
	<Segment StartTime="5013.601" EndTime="5014.160" Participant="me034">
	Um.
	</Segment>
	<Segment StartTime="5018.486" EndTime="5022.466" Participant="me013">
	just translate to articulatory features, some set of articulatory features and train with <Emphasis> that. </Emphasis>
	</Segment>
	<Segment StartTime="5023.230" EndTime="5030.125" Participant="me013">
	Now the fact - even though it's a smaller number, <Pause/> it's still fine because you have the - the, uh, combinations.
	</Segment>
	<Segment StartTime="5030.980" EndTime="5035.716" Participant="me013">
	So, in fact, it has every, you know - it had - has - has every distinction in it that you would have the other way.
	</Segment>
	<Segment StartTime="5035.248" EndTime="5035.686" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="5035.854" EndTime="5037.980" Participant="me013">
	<VocalSound Description="inbreath"/> But it should go across languages better.
	</Segment>
	<Segment StartTime="5038.620" EndTime="5040.865" Participant="me018">
	We could do an interesting cheating experiment with that too.
	</Segment>
	<Segment StartTime="5041.249" EndTime="5041.705" Participant="me018">
	We could -
	</Segment>
	<Segment StartTime="5042.550" EndTime="5043.423" Participant="me018">
	I don't know, if you had
	</Segment>
	<Segment StartTime="5044.194" EndTime="5044.798" Participant="me018">
	uh
	</Segment>
	<Segment StartTime="5047.316" EndTime="5050.820" Participant="me018">
	the phone labels, you could replace them by their articulatory features and then feed in
	</Segment>
	<Segment StartTime="5051.211" EndTime="5052.613" Participant="me018">
	a vector with
	</Segment>
	<Segment StartTime="5053.028" EndTime="5053.558" Participant="me018">
	those
	</Segment>
	<Segment StartTime="5054.253" EndTime="5056.461" Participant="me018">
	uh, things turned on based on what
	</Segment>
	<Segment StartTime="5056.743" EndTime="5062.813" Participant="me018">
	they're supposed to be for each phone to see if it - if you get a big win. Do you know what I'm saying?
	</Segment>
	<Segment StartTime="5063.403" EndTime="5063.941" Participant="me013">
	No.
	</Segment>
	<Segment StartTime="5064.420" EndTime="5068.293" Participant="me018">
	So, um, I mean, if your net is gonna be outputting,
	</Segment>
	<Segment StartTime="5068.293" EndTime="5075.139" Participant="me018">
	uh, a vector of - basically of - well, it's gonna have probabilities, but let's say that they were ones and zeros,
	</Segment>
	<Segment StartTime="5075.139" EndTime="5077.996" Participant="me018">
	then y- and you know for each,
	</Segment>
	<Segment StartTime="5077.996" EndTime="5082.466" Participant="me018">
	um, I don't know if you know this for your testing data, but if you know for your test data,
	</Segment>
	<Segment StartTime="5082.466" EndTime="5087.058" Participant="me018">
	you know, what the string of phones is and - and you have them aligned, then you can just -
	</Segment>
	<Segment StartTime="5087.058" EndTime="5093.506" Participant="me018">
	instead of going through the net, just create the vector for each phone and feed <Emphasis> that </Emphasis> in
	</Segment>
	<Segment StartTime="5094.220" EndTime="5094.894" Participant="me018">
	to see
	</Segment>
	<Segment StartTime="5097.202" EndTime="5103.390" Participant="me018">
	if that data helps. Eh, eh, what made me think about this is, I was talking with Hynek and he said that there was a guy at A_T_and_T who spent
	</Segment>
	<Segment StartTime="5103.390" EndTime="5106.200" Participant="me018">
	eighteen months working on a single feature.
	</Segment>
	<Segment StartTime="5106.650" EndTime="5109.144" Participant="me018">
	And because they had done some cheating experiments -
	</Segment>
	<Segment StartTime="5107.501" EndTime="5108.141" Participant="me013">
	<VocalSound Description="inbreath"/>
	</Segment>
	<Segment StartTime="5108.736" EndTime="5115.295" Participant="me013">
	This was the guy that we were just <Emphasis> talking </Emphasis> a- that we saw on <Emphasis> campus. </Emphasis> So, this was Larry Saul who did this - did this. He used <Emphasis> sonorants. </Emphasis>
	</Segment>
	<Segment StartTime="5113.149" EndTime="5113.994" Participant="me018">
	Oh, O_K.
	</Segment>
	<Segment StartTime="5115.600" EndTime="5117.011" Participant="me018">
	Right, O_K, right.
	</Segment>
	<Segment StartTime="5116.062" EndTime="5117.011" Participant="me013">
	Was what he was doing.
	</Segment>
	<Segment StartTime="5117.011" EndTime="5117.461" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="5117.011" EndTime="5120.076" Participant="me018">
	And they - they had done a cheating experiment or something, right? and determined that -
	</Segment>
	<Segment StartTime="5118.961" EndTime="5121.088" Participant="me013">
	He - he di- he didn't mention <Emphasis> that </Emphasis> part. But.
	</Segment>
	<Segment StartTime="5121.088" EndTime="5128.009" Participant="me018">
	<Uncertain> Well, </Uncertain> Hynek said that - that, I guess before they had him <Emphasis> work </Emphasis> on this, they had done some experiment where if they could get that one feature right,
	</Segment>
	<Segment StartTime="5125.012" EndTime="5125.716" Participant="me013">
	I see.
	</Segment>
	<Segment StartTime="5127.903" EndTime="5129.061" Participant="me013">
	O_K. <Comment Description="very quietly"/>
	</Segment>
	<Segment StartTime="5128.009" EndTime="5129.433" Participant="me018">
	it dramatically improved
	</Segment>
	<Segment StartTime="5130.090" EndTime="5133.745" Participant="me018">
	the result. So I was thinking, you know - it made me think about <Emphasis> this, </Emphasis> that if -
	</Segment>
	<Segment StartTime="5133.745" EndTime="5137.020" Participant="me018">
	it'd be <Uncertain> an </Uncertain> interesting experiment just to see, you know, if you <Emphasis> did </Emphasis> get all of those right.
	</Segment>
	<Segment StartTime="5137.630" EndTime="5142.659" Participant="me013">
	<Emphasis> Should </Emphasis> be. Because if you get <Emphasis> all </Emphasis> of them in there, that defines all of the <Emphasis> phones. </Emphasis> So that's - that's equivalent to saying that you've got -
	</Segment>
	<Segment StartTime="5138.274" EndTime="5139.275" Participant="me018">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="5141.959" EndTime="5142.418" Participant="me018">
	Right.
	</Segment>
	<Segment StartTime="5142.659" EndTime="5144.016" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="5144.016" EndTime="5146.495" Participant="me013">
	got all the <Emphasis> phones </Emphasis> right. So, if that <Emphasis> doesn't </Emphasis> help, there's - <Comment Description="abandoned as clear"/>
	</Segment>
	<Segment StartTime="5146.327" EndTime="5147.174" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="5146.495" EndTime="5152.425" Participant="me013">
	Although, yeah, it <Emphasis> would </Emphasis> be - make an interesting cheating experiment because we are using it in this funny way, where we're converting it into features.
	</Segment>
	<Segment StartTime="5150.718" EndTime="5151.410" Participant="me018">
	Yeah.
	</Segment>
	<Segment StartTime="5152.361" EndTime="5156.073" Participant="me018">
	And then you also don't know what error they've got on the H_T_K side.
	</Segment>
	<Segment StartTime="5152.425" EndTime="5153.663" Participant="me013">
	<VocalSound Description="laugh"/>
	</Segment>
	<Segment StartTime="5156.520" EndTime="5156.886" Participant="me018">
	You know?
	</Segment>
	<Segment StartTime="5157.511" EndTime="5157.890" Participant="me013">
	Yeah.
	</Segment>
	<Segment StartTime="5157.890" EndTime="5160.978" Participant="me018">
	It sort of gives you your - the best you could hope for, kind of.
	</Segment>
	<Segment StartTime="5161.199" EndTime="5161.503" Participant="me006">
	Mmm.
	</Segment>
	<Segment StartTime="5163.290" EndTime="5164.813" Participant="me006">
	Mmm, I see.
	</Segment>
	<Segment StartTime="5165.200" EndTime="5169.488" Participant="me034">
	The soft training of the nets still requires the vector to sum to one, though, right?
	</Segment>
	<Segment StartTime="5168.333" EndTime="5170.270" Participant="me006">
	To sum up to one.
	</Segment>
	<Segment StartTime="5169.488" EndTime="5174.760" Participant="me034">
	So you can't really feed it, like, two articulatory features that are on at the same time
	</Segment>
	<Segment StartTime="5174.760" EndTime="5176.374" Participant="me034">
	with ones
	</Segment>
	<Segment StartTime="5176.374" EndTime="5179.678" Participant="me034">
	cuz it'll kind of normalize them down to one half or something like that, for instance.
	</Segment>
	<Segment StartTime="5179.678" EndTime="5184.962" Participant="mn007">
	But perhaps you have the choice of the <Pause/> final nonl- uh, nonlinearity, yeah.
	</Segment>
	<Segment StartTime="5180.324" EndTime="5181.332" Participant="me006">
	Right.
	</Segment>
	<Segment StartTime="5182.994" EndTime="5184.034" Participant="me006">
	Nonlinearity?
	</Segment>
	<Segment StartTime="5184.743" EndTime="5189.147" Participant="me006">
	Um, it's sig- No, it's actually sigmoid-X_
	</Segment>
	<Segment StartTime="5184.962" EndTime="5187.737" Participant="mn007">
	Is it always softmax or - ? <Comment Description="leading question"/>
	</Segment>
	<Segment StartTime="5188.715" EndTime="5189.348" Participant="mn007">
	Yeah.
	</Segment>
	<Segment StartTime="5189.851" EndTime="5190.785" CloseMic="false">
	<NonVocalSound Description="door squeak"/>
	</Segment>
	<Segment StartTime="5189.874" EndTime="5190.775" Participant="me006">
	for the -
	</Segment>
	<Segment StartTime="5191.377" EndTime="5194.591" Participant="mn007">
	So if you choose sigmoid it's o- it's O_K?
	</Segment>
	<Segment StartTime="5193.970" EndTime="5197.758" Participant="me006">
	You, um - I think - I think apparently, the, uh -
	</Segment>
	<Segment StartTime="5194.299" EndTime="5196.170" Participant="me013">
	Did we just run out of disk, or - ?
	</Segment>
	<Segment StartTime="5196.869" EndTime="5198.017" Participant="me034">
	Why don't you just choose linear?
	</Segment>
	<Segment StartTime="5198.240" EndTime="5198.716" Participant="me006">
	What's that?
	</Segment>
	<Segment StartTime="5198.565" EndTime="5199.731" Participant="me034">
	Right? Linear outputs?
	</Segment>
	<Segment StartTime="5199.855" EndTime="5201.016" Participant="me006">
	Linear outputs?
	</Segment>
	<Segment StartTime="5200.960" EndTime="5201.776" Participant="me034">
	Isn't that what you'll want?
	</Segment>
	<Segment StartTime="5201.776" EndTime="5202.780" Participant="me006">
	Um.
	</Segment>
	<Segment StartTime="5201.783" EndTime="5202.780" CloseMic="false">
	<NonVocalSound Description="scratch noise on mike"/>
	</Segment>
	<Segment StartTime="5203.490" EndTime="5205.274" Participant="me034">
	If you're gonna do a K_L Transform on it.
	</Segment>
	<Segment StartTime="5206.774" EndTime="5207.830" Participant="me006">
	Right, right.
	</Segment>
	<Segment StartTime="5208.658" EndTime="5210.248" Participant="me006">
	Right, but during the training,
	</Segment>
	<Segment StartTime="5210.248" EndTime="5213.404" Participant="me006">
	we <Uncertain> would train </Uncertain> on sigmoid-X_ and then
	</Segment>
	<Segment StartTime="5211.147" EndTime="5211.868" Participant="me034">
	<Uncertain> Oh, you - Yeah? </Uncertain>
	</Segment>
	<Segment StartTime="5213.404" EndTime="5215.983" Participant="me006">
	at the end just chop off the
	</Segment>
	<Segment StartTime="5215.983" EndTime="5217.462" Participant="me006">
	final nonlinearity.
	</Segment>
	<Segment StartTime="5216.075" EndTime="5216.873" Participant="me034">
	Hmm.
	</Segment>
	<Segment StartTime="5219.060" EndTime="5221.270" Participant="me013">
	So, we're - we're - we're off the air, or - ?
	</Segment>
	<Segment StartTime="5221.179" EndTime="5222.029" Participant="me011" CloseMic="false">
	Um, about to be <Comment Description="unmiked, audible in the background of [me018]'s channel"/>
	</Segment>
	<Segment StartTime="5222.140" EndTime="5222.894" Participant="me013">
	<Emphasis> About </Emphasis> to be off the air.
	</Segment>

	</Transcript>

	</Meeting>