me013 left early, and did not read digits. Is it on? O_K, here we go. Uh, so - uh, so I c- I definitely have to leave at three-thirty today for whatever. That's - So - Definitely have to go to tea. So - Yeah, they're - they're having birthday cake and such and so I can't miss that. That too, but - You're a May birthday, right? I'm a May birthday, yeah, so I have to - Oops. Barbara told me. Barbara Peskin. Yeah. Um - So, only a couple of agenda items, since no one sent me email for agenda items. Uh, the first is the I_B_M transcripts. I'd like to uh, mention - I s- I spoke with several of you about this, but just to brainstorm a little bit about what we can do with it. Mm-hmm. So we got back the I_B_M transcript, and the accuracy looks O_K. There are - I mean, there are areas which are clearly wrong and there are a lot of areas where they put question marks, where the acoustics weren't very good. And I think that's fine. Putting it in question marks is better than doing it wrong. The big problem was, they had the wrong number of beeps in it. Oops. So it didn't align. And that's exactly what I was afraid of. And I went in, and more or less by hand, corrected it, by loading it - I wrote a script that will convert it to the multi- the Channeltrans format, looked at it in - in Channeltrans and found the places where it got a- unsynchronized, and then either - Hmm. Inserted ones. And in this case only - only had to remove beeps, because that was th- always the error, is that they put in it extraneous beeps. But it's not easy to do, Hmm. because the whole thing gets offset, and basically, it's assigning the wrong speaker to the text, and it all gets unaligned. So it's very difficult Hmm. to do. So it took me several hours So this is - just to do that. For one meeting. this is one - it's one hour of - of - of speech? Yep. And it took you a few hours to - to do it. It took me a couple hours to write the scripts, and then it - it probably took me maybe forty-five minutes, just going through it. Now obviously it will be a lot faster from now on, because now I sort of understand how it works. But, I think if we had some way - uh some tool that made it easier to insert and delete beeps it would make it a lot easier. Cuz one of the problems is that it takes a very long time when you make a change to load it back up in the channel transcriber. Yep. Yeah, that's true. I've noticed that. Yeah. One of the things that Adam said that I thought would be a good idea would be, um, And so there may be - Which makes sense. if we g- used different beeps. So for example if we alternated between two beeps, a high and a low. He said that? th- Uh, well in - in some email that, uh, we had. Oh, on the email. For - Oh, oh, O_K. Yeah. I didn't just say that, no. Yeah. Um - Because then - So th- so this is the o- the other approach which is, what do we do in the future, to try to make this less common? But it's still gonna happen. Yeah. Well, and then the other possibility was - e- that we provide them with a - a file that already has the beeps in it or something. Wasn't that what you said? So - Hunh. Sort of a text template, but I don't know how Uh-huh. Yeah, we don't know what their process is. So I had two ideas. The first was to provide them a text template that had both the beeps in it and the speaker I_Ds. we can do that. Yeah. You know, just "male-female", "English-nonEnglish", "one two three four". Um - Can I - I - I just - And they just filled it in. How- how many - how many beeps are there, and how many do we- were they - d- were - were off? It was like a hundred twenty, and they had a hundred twenty-three or something like that. Is that right? Alright. Well, l- I- I thought it was a lot more than that. And is it a - Well I - I don't remember. They were three off, basically. Yeah, and it was a hundred-ish. And this is the mixed signal? Oh! Hmm. So - so if you get one - uh, uh, excuse me. If you get one off, it - t- everything is off cuz of the alignment right? Yep. Yep. Uh, and this is all in one file that's continuous. Yep. Yeah. So why don't you make it multiple files? How would that help? Because then if they have one off, then - then it would only screw up that file. Oh, you mean just break the one hour into Yeah. a few chunks. I mean, it seems now, if you just ha- make one mistake then everything is off. Well the only - the only question about that that I have is, um, I think Brian has to put all this stuff onto a tape. So I don't know whether - Yeah. I'm sure they would prefer it sort of one hour. Can he send them - I guess he could send them multiple Oh! tapes Uh - Yeah, just say, "O_K, is -" ten minutes is -" I mean, maybe there's other things to do that are more clever, but this just seemed to be the sort of the obvious thing because, again, if you make one mistake, then everything after that mistake is off. Mm-hmm. And, uh - Well, the only other issue is we'd have to worry about reassembling them in the proper order. Uh - Yeah. That - I mean, just naming schemes will get - We can definitely do that. So t- so one question is, are there good places in the files where we can really do that? Well, the other thing - You can - I mean, so - so - so someone will have to go through them and listen to them and pick a place to break them. Yeah. The other thing you can do- This is on the mixed signal, or are these the individual - the ones with the beeps that they're getting, that they made mistakes on is They are played from the the mixed one? Individual channels. Individual channels. individual. Or they're indivi- If it's individual, um - y- c- we can - probably run a forced alignment, where you c- include the beeps as words and pretty much figure out where the errors are. I don't think we could do it on the mixed signal, cuz it - But on the individual channels that would probably work, if they have enough words in the transcript. Yeah, we - In fact, that's - Yeah, maybe we - So, I mean, that's pretty quick to run. It - it's, um - I mean, we could try it. Cuz if a beep is treated as a You mean, so, actually put - put the wavefile in with beep in it. unique word, "beep" - Put in - put "beep" in as a word. And pu- and have a beep model. We have the original wavefile with all the beeps. Yeah. Yeah. Well - the beep model - Actually, you could train a beep model. That - that's a good idea. It would be very good. Um - Yes, it would be, cuz it's very, very regular. Yeah. Yeah. So, um - Oh. Kind of low variance, wouldn't that be? You sort of might end up with Yeah. Yeah, very low. Yeah. Um, that might actually - some infinities there, but - You have to - may - add some jitter to it. Can I see that? I mean, that might actually work, because you'll - Where's the case? I guess the trick is figuring out, you know, where it doesn't align? But if it's a problem of extra beeps, rather than missing beeps, then - um - that's easier It could t- it could - and it might work. Oh. There could be some missing ones. But - So is this a It's just in this case they were not. seven or something? Or - What is it? M_one-hundred. So I think - uh, Chuck had a good explanation for it which I liked, which is these - they're listening to it and so they write something down, and let's say they miss it, they rewind. M_one-hundred. Oh. Well, they hear the beep again. Was that that beep, or was that a different beep? So that's why Chuck was suggesting different tones. My original thought Oh, yeah. was, you could have a different tone for each speaker. You know, there's plenty of tone space. And that might help cue the transcriptionist. You know, give them a little cue about where they are. Or you could - you could have ascending tones, or something. What do you mean? Uh - Or - Yeah. Increasingly high - higher frequencies. You could have, um, each speaker in the meeting say "beep" and record it and that will be their pro- No, I just - thi- I don't know. If you have a hundred or something just - just sort of move them up i- mu- up a - a half tone or something for - for each one, and then - Sorry. That's what I was saying. Is that - that you could have - each speaker could have their own tone. Oh was it? Oh. See, I should listen. Yeah? Hmm. And you know, since there are no more than, you know, ten speakers or so per meeting, Oh. Well - there's plenty of tone space. But - Uh, O_K, yeah. So the only - That might work too. Yeah, I was just thinking that actually temporally just, uh, uh, uh, an increasing tones so that Oh, O_K. they would really have a sense of whether they were going, you know, forward or backward or - Yeah. Bo-eep be-oop. In the end? Yeah. Well you th- So just r- r- truly chron- chromatically up all the way through the entire meeting. Yeah. Well, it's a thought. Interesting. Oh, oh, th- oh, I see. It'd be impossible to r- to misorder them. Yeah. Yeah. But yo- I guess you'd still get the mistake that you mentioned, because the person - I mean, if that's some of the errors, those would probably stay the same, regardless. Well - Yep. I mean, Chuck's suggestion of just two beeps is nice because then you could have them actually transcribe "H_beep" or "L_beep", "high beep" or "low beep". Yeah, maybe. And then when we get them back, if we see two L_beeps in a row - It would be much easier. That's probably better, yeah. S- scratch my other idea. Yeah, you have like a sanity - sanity check on them. That's a good idea. Alternating. It's pretty simple, pretty simple. Yeah. Yep. Yeah. And it would - it seems like it would help them, you know, if they just finished a long passage and they heard one beep and then they hear it again, they know it's the same beep. Versus if they heard a slightly different one. Yeah. Hmm, interesting. Yeah, that's a good, like, error checking approach, I think. Yeah, we can, yeah, like, do - do, uh, one for each ca- one for each person in the meeting, you can only hire people who have perfect pitch, and they can say "A_flat" "B_". Well. Well, "high", "low" is pretty straightforward, yeah. Great idea, yeah. No, two is probably better than what I was suggesting, because what I was suggesting, in order to make it through a reasonable range for the whole thing, you'd have to have the - the tonal differences between the two kind of small, and actually you want them large. So. Yep. You'd have to - they'd have to have the acoustics of dogs t- hear into the high high ranges. Mm-hmm. Yeah. Yeah. So I think, certainly, doing two beeps is - is no-brainer. And then the other question is, if they can - If they do something on a computer in a format we can handle, we could give them a text file that was a template with speaker I_D and beeps already in it. Yeah. And then they could just fill that in. Maybe I should write to Brian and tell him what the problem was and what our proposed solutions are. So what - Right. See what he thinks would work best with them. Yeah. I mean, what's going wrong on the other end? What are they doing? What, uh - Well, as I said, I think - We don't actually know, but I think Chuck's hypothesis was a good one, which is, you re- r- you listen to something, you write down what you thought you heard, but you want to listen to it again, so you r- rewind. And there are some - And then you hear the beep again, and then you say "well, is that that beep, or is that a new beep? I don't remember." And so, you know, a couple times they got it wrong. And there are some cases where there's, um, very little speech before you hear a beep. So you hear a little bit of speech, beep, a little bit of speech, beep. Yeah. Mm-hmm. Maybe three of those in a row, and was that three beeps, was that two beeps? Um - See, they have th- Brian said the setup that they have is, um - uh - i- um - They've got headphones and then I guess they have a computer, and they have a foot pedal which lets them quickly scan back and forth through the tape. I- i- t- i- s- makes the tape go real fast forward or fast backward. And so they can hear something, then step on the pedal and quickly rewind and go back and hear it again. And so they're - I think they're probably going over and over f- sections. And then they get confused about whether they've already put that beep, or if they heard, you know, three beeps in a row, was that three or two, and - It's a shame we can't really give a number. So it would be a number and a beep. O_K, s- e- It's a shame we can't do that. Well, we could. Oh, that's an interesting idea. We could, but then that starts getting pretty long. No, no, no. But wait, maybe - maybe what we have instead of a beep is a, uh, synthesized number. Yep. And they put the number in when they hear it. That'll work. So, inst- replacing the beep with "twenty- three ", "twenty- four ", "twenty-five" - And then they have to transcribe what that number is each time. Wow, that'd also be shorter to type. Yeah, just a "two three, two four", that would be - Well, they would have to have a mark. Interesting. There'd be some kind of mark in front of it. Well, cuz they're s- distinguishing between twenty-three - that's your indicator there, and twenty-three that somebody happens to say in a meeting. Oh, that's true. Well it would have t- yeah, it would have to be um, you know either obviously a s- sounding synthesized, uh, little - Yeah. But they might miss that, right? Or maybe could it be a click and the number. Yeah. Or a beep and the number. Well, it's sequential. Or a beep and the number, and then - and then you just do the - the num- th- you do it as a d- a digit. Right, you would go sequential. So unless you got pretty unlucky, Uh, right. Yeah. what the person was saying and the number - In fact, we could put it at the beginning of each utterance. We could say "beep one ". And then they'll hear the speech. I think having a beep too is good. So beep number. Yeah the- It's just getting pretty long. You know, the utterances are very short. Or a number beep - Yep. Mm-hmm. And so you're gonna be talking beep, number, "yes!" beep, number, "no". Mm-hmm. Well, if - if it saves time, yeah. Yeah, and u- I think it would - I - Actually, I thi- I like that. I think it would - And when they transcribe this meeting it's gonna be really impossible. Yeah, that's funny. W- we're saying "beep". Yeah - for instance. Yeah, yep. Yeah. But see that would - that would definitely keep - It's not a bad idea. Yeah. Keep things from getting out of sequence. If they heard it over again they would know. It would be obvious from their transcript whether they did that one or not. It might actually help them in terms of their place holding. Yeah. That would help get them synchronized. Yeah. And it might be helpful to them in terms of d- act- I think it would be, because they would know their place. Yeah. And we have plenty of digits data, so. They would know their place, darn it! We- Know - Those transcriptionists need to know their place. Oh. Well, no no no no. But it sounds like a k- a - a - a short conversation with Brian might be - might be helpful to get a sense of what's going on. Yeah. I'll - I'll - I'll write him an - Yeah, the numbers are a good idea. O_K. Yeah. Or just - Othe- other than lengthening the transcript, I think it would be very helpful. Just instead of the sequential number, for- perhaps just the speaker I_D or something. Just say speaker one or spea- or just one, two, three, four for four speakers or something. Pppt! I - i- Could - could be helpful. Yeah, you don't have to - you could keep them short by not s- or just go one through ten, one through ten, one through ten, or one through twenty. Yeah. Yeah. Yeah. Yep. And then also, if we - I mean, I don't know what their setup is, but your template idea might be a way of - of Mm-hmm. of - We could actually build the numbers into a template. Yep. Well. I'll talk to Brian, and see what he - Yeah, single digit numbers - Yeah. That also appeals to me. So that you don't have to, you know - That's a good idea. Very simple. "One hundred twenty-four" Yeah. Yeah. Well, th- although - "Beep. One thousand three hundred forty-two" You're saying i- it's - it's about an hour meeting, and there were a hundred and twenty-three - Mm-hmm. I don't remember. It's forty-five minutes. Chuck was saying there was more than that. Hmm? Forty-five minutes for that one. More beeps than that. I was thinking there was on the order of fifteen hundred segments, but maybe I'm - There must be. Because otherwise, you're going a whole - I think it should be more, yeah. It's possible. That was at least forty-five minutes, right? Yeah. That's right. Was it that many? I just don't recall. Yeah, it was forty-five minutes. That's the shortest meeting we have. Most of them are double that, or Yep. Well, I was just wondering what the average length so Oh, of a - of a segment that they're hearing? thing was. It's - Yeah. I don't know. Hmm, interesting. I don't know. Uh- Easy enough to figure out. We have all that data. Yeah. So if there are in the - in the hundreds and thou- or thousands of these - It must be in the high hundreds of them, at least. So, right. So i- So that's getting a little cumbersome for them to type, like, Yeah. Well, with just one digit though, five hundred? I think - Well - But yeah, if they recycle through - Yeah, if we - if we did mod ten, then - Yeah. Right, but my point was, you know, if - It's a big difference to me if there's a thousand or there's a hundred. If there's a thousand, then that means that this is only happening three times out of a thousand. Which means, you know, even just splitting it up into a few files would probably take - You know, nearly all the time, there would be no - Yeah. W- if it's a hundred, then - you know, that's not the issue, but if it's a hundred, then, the average length of each one is pretty - pretty long and so having a little thing at the front of it is no big deal. Mm-hmm. So, that's why I was curious. Yeah, you're right. It had to have been more than that. I mean, it was a forty-five minute meeting, so - Right. And it certainly was not a minute a chunk, it was a few seconds a chunk. That's what I - Mm-hmm. Yeah, so it'd be a lot of overhead to type. So I'm just mis-remembering. I'm just mis-remembering. Oh, I - So. Well, and the other thing is, maybe the numbers would help them keep their place in the transcript and speed the process - Yeah. It might make the transcript faster - W- wel- I mean I - A quick conversation with Brian would be good, so. Yeah. Mm-hmm. Yeah. O_K. And other than that - Actually, if it is - u- I mean, it's not a bad idea to try the alignment. Um, if there's only like three of them, then if you align, the first point at which things get messed up is sort of the location of the first problem. Mm-hmm. And then, the second p- I mean, you could do it that way, or you could - It's - it's - it would be able to handle any errors. So if they make, on the average, that many, Right. and it costs them more time to do something different than what they're doing, which I don't know, but if it does, then we could try doing that as a post-process, and, um, have a student or a transcriber run this alignment, or we can do it and - Yep. You know actually, what - all that we would - then you can iteratively figure out where the problems are. You know, Liz, all we would re- It would take a little work, but not any real human - not a lot of human work. Right. I mean, that's what we want to avoid. I mean if - in terms of the alignment, actually, all we would need is - we wouldn't even need words. We would need a general speech model and a beep model. Cuz all we're really concerned about is missing beeps or too many beeps. Well, the for- the forced alignment will run fa- otherwise it'll take forever. I mean, So if we had a - Mm-hmm. to really run recognition? Well - l- I guess what I'm thinking of is that w- Or - ? that a lot of times there will be - they'll put in question marks, that represent some unknown amount of speech, so we're gonna have to have some kind of - Right. Yeah, then we just map it to a reject model. In - and in fact, that's what we do now. Cuz there's cases, even after Jane listens, where, you know, we have - Right, h- an expert can't even tell what they're talking about. Yeah. Um, so that's O_K. Just map that to reject and - Right. Well, what I was thinking is, map all the speech to the reject model. And then you have a beep model. Yeah. But that won't work because it's - Yeah, you're still talking a forced al- The forced alignment might not work then, although we can try it. Um - Well, but, all we'll get - Well, with the beep models, is what's really - you'd have really g- it would be really good though. So - The beep model is - is gonna match, like, really well. Right. so it's p- so it's really just doing, yeah, a keyword, like a spotting for beep with - Right. But it'll - it'll grab the next beep, in other words. It'll be - You'll get back another offset. Um - I mean, I was thinking of - then I - and then I realized "Well, the recognizer will just go along and line up all the beeps, and then there'll be all the extra beeps at the end if there were more beeps then you wanted." Right, you - you need - But you need the text So you need the word - you need the word to c- sort of control the relative location. to tell it where it got it wrong. Yeah. No, I'm not sug- I'm not suggesting that we don't have anything between the beeps. It will tell you - Oh. But I - what I'm saying is, what's between the beeps we don't really care about. I think you do care about it. Because you have to know which ones match which. Unless you wanna know if they're - Right. Otherwise you just count up the total And we can do that with wordcount. number of beeps. Right? We already know that - that we're three short. And so if we just did Yeah. spee- uh beep and non-beep all you would get is that you have three mistakes, but you wouldn't tell you which ones were wrong. Right. You need some of the words in there, so that you can say "well, this segment matches this one, this segment matches this one, and this one doesn't match at all, unless we insert a beep ". I mean the only thing I'm worried about with that approach is that if O_K. we need to figure out the beep alignment problem before the transcribers do the corrections here, then we're in trouble. In other words, if the transcripts aren't sort of good enough that the aligner constraints are good enough to sort of show you where the errors are, then it wouldn't work. But it might work. It might work to do this if their transcripts are pretty close on the words that usually get recognized correctly, which are the, you know, function words, the common words. Well, actually, you know, there's an even easier way. We don't really need a beep model. Um, just extract the segments and do a forced alignment, Right, you can also - and if the score is good, then you say it matches - In fact that's what we do - Right. The individual segments between the beeps - Um, if beeps were like the segments that we get from um, the transcription tool already, that that's what we have done, and it works very well. You can see these segments align and these don't. Um, then you just have to go back in and figure out where the endpoints of those segments are, cuz some of them will be wrong. Because the bi- the beeps were missing, by definition. Right. So do you get a score for each alignment? Is that what you get? So it might actually work. Um, you ca- you get, um - Definitely when they don't align at all, it - it - it fails. I mean, that - that's how we found a lot of Mm-hmm. problems before, with, um, words being on the wrong place or something. So a failed alignment is a very good indicator that - Mm-hmm. that - the words don't match up. I'm just wondering when - wha- what it means to fail. Um - Is it the - the likelihood scores below some threshold - Yeah, is, uh - it - Right. It can't match up the - That's why you actually need the text. In order to force you to try to match something that gives you a model to match against. So - It's just an idea, if - if it turns out that - I mean, I also like this idea of high and low beeps, or - But it - if - Suppose we get one or two errors, still, per - you know, per transcript, then we z- we might wanna try some approach like that. Well, the numbered ones would make it a lot easier, cuz you could then really localize where the error is. Mm-hmm. Yeah. Where it is. Right. Yeah. If - so, if that doesn't add time for them, that'd be great. You know, like - Yep. I think it would save them time. I think it would help them keep the track - track of it. We have to ask them. Uh - uh, again, especially if there's hundred - many hundreds or thousands I still think it would - it would cut the incidence of this a lot if it was possible to break it into a couple files. Interesting, yeah. Yeah. Cuz it, you know, it's r- r- I guess also even for the transcribers, it's quicker to load a smaller file - into the - You know, t- for the checking problem. Yeah, I mean, we could certainly break the meetings - Into We could certainly break the meetings into pi- pieces. twenty minutes chunks or something, or - So, just - I mean, you've already broken them into chunks, so if you have - if you have a th- if you have a thousand chunks then Just - Just accumulate it till you get twenty minutes worth? make ten things of a hundred, or - or you know. There's one file. That's true, yeah. Yep. And then it will be, you know, this - Eve- every now and then there'll be a - a beep missing. So, someone's gonna talk to Brian. Yeah, I'll - I think that was Chuck. I'll talk to him. O_K. Yep. I also looked thr- over the text and I was impressed by the accuracy, it looks really good. Yeah, it does look good. I mean, I found several errors, but they were not significant. They were all things that I could easily listen to and sometimes convince myself they said one thing and sometimes the other, so. Yeah, it's there. Mm-hmm. I haven't looked at these, um - cuz I was gone last week, but - but Don had told me that there's a difference in some of the conventions? Oh. Yes. But this is not important. It's systematic, yeah. So - but those are all easy things, right? Yeah. O_K. And - and Brian actually forwarded me in advance. He - he wor- he very nicely worked out a set of conventions which is intermediate between ours and the ones that they're used to. Which is - which is really a good - a good way to do it. O_K. So we can just map - map them to - No problem. Great. Hmm. O_K. Um - We still haven't really sat down and talked about file reorganization, and directory reorganization. So we still have to do that. But I don't think we need to do that in this meeting. But, uh - It is something that needs to get done, and I wanna also coordinate it with Dave so that we do a level zero backup right after. So we don't waste a lot of tapes. But, uh - So - let's - let's try to do that sometime, O_K? ch- O_K? Yeah. We can do it this week. Yeah. Um - We also still have to make a decision about mike issues, what we wanna do with that. I thought we were going to get two more of the ones - And just swap them in and out? Yeah, we could certainly do that. So why not? So - uh, Morgan, just to - uh, since you weren't at the meeting last week, uh, apparently a bunch of the E_D_U fe- folks really hate this style mike. Um, I wouldn't - well, it - they didn't say "hate" but they - they c- they come on time to their meetings in order to not be left the last person who has to sit by those Can I - I c- I - I mean, that's - that's - that's a fairly strong indication of dislike. mikes. I actually really like them though. Um - Which one is these? The uh - There were a few people - you, and like, three or four people who really like them. The Crown. Uh-huh. Yeah, I do. And e- and th- all the others really don't. So what I was thinking is, we could get a few more of the Sony ones and just unplug them and plug them in. And the only thing is that when you fill out the digit forms, you have to be sure to indicate which mike was actually used. So, I mean, that's easy to do. Um - it - it - moves us away from this uniform, all the mikes are the same, which some people had said was a benefit. But - The other thing, too, is th- y- you said that the acoustic pra- q- quality of the one that you're wearing, the fancy one, is actually somewhat better than It is. Yeah. So I- I don't know if it's a noticeable difference, but it is - I mean, I p- I personally really prefer that model, so I think you know we've got this problem of people having different - different preferences. You know, there are so many things about this that are non-uniform that I think having - having some modicum of uniformity is a good thing, but you're just not going to - I mean we've - O_K. uh - For instance, jus- just the fact that people sit in different positions, different times, the - exactly where they put it with respect to their mouth is different each time. Yep. Uh - I - I just don't - I mean, I think we - we're - we're getting rid of major differences like we're not using the lapel anymore, an- and so forth. But Mm-hmm. the other stuff is just gonna - some variability we're gonna have to accept. It's - I - I understand, that's fine with me. I - I g- would say that um, I find this model easier to adjust, that's one of the reasons I don't like these, cuz I keep m- Uh-huh. you know, bending it, and then it's n- and I never really know - Yeah, see I - I happen to like the old one better. You do? O_K. But - Yeah. What about just a different headband thing? And some people do, but - Or even if we could attach two headband - Like, I don't mind those, but it - I mean - th- We abs- th- We have lots o- of choices of microphones. So - they bounce around. I mean, can we keep the microphones, and just Mm-hmm. somehow attach a more comfortable thing over your head? Mmm. Yeah. That's my problem with this one. The - the ear thing comes out to here. It doesn't even fit over my ear. Right. Hmm. And for some of the others, it's - their ears were shaped in a way that didn't hold the - Right. @@ Huh. I mean, so we can go microphone shopping, and get - get more microphones. It's just - w- Then they have to be rewired though, don't they? I'm just not sure what we should do. Yeah. You know. I mean, here's another argument against worrying too much about uniformity. We have something like forty-five hours of data collected. Specially. And, uh, we're also gonna be integrating in this corpus stuff that's - that's being done at U_W, Yep. uh, with entirely different microphones. So I - I think, you know, getting away from the lapel, having close-mounted mikes, having people adjust them as they're most - as - as - as comfortable as they can - Well, I - I think that's sort of the big deal. I agree with that, and so my feeling is, we should get mikes that people like. Yeah. Yeah. And so I- my feeling is, going out and buying a few more mikes is fine. Yeah. Yeah, that's right. So should I just go do that? Yeah. O_K. What about these - And then also, should we go ahead and get another wireless system? You know, for whatever it's gonna be, uh - Mm-hmm. Uh, another wireless system. You mean, another transmitter? Or - or - Uh, no, actually, get - we need another box. Because each - each box in the back room can only take six. Oh, to get r- to get rid of the wired things. Yes, absolutely. So we could- Yep. O_K. Yeah. Uh, because - cuz we've had - With the exception of forgetting to change the batteries from time to time, uh, we've really had no tr- no trouble with this at all, right? And - So I'll just go d- I'll do that. Yep. and the wired stuff is just gonna continue to degrade as, you know, and nobody will remember how to - Another one just bit the dust. yeah, nobody will remember how to repair a Jimbox or a Jimlet or a - you know, it's - Really? I think there's a problem with this one. Yep. O_K. Then I will just go do that, and send the bill to - Yeah. uh, what's-her-name. Whoever will pay, yeah. Whoever will pay. You. Yeah, yeah. I have a quick question about microphones, um - I got - this crazy idea that, um, i- in the future, people will just walk around with the microphones that they use for their cell phones? You know, these little boom ones, like, and really go to meetings with close talking mikes. If they're their own personal microphone. And so I'm wondering if we can get a couple of tho- I don't know how good quality they are, but it would be really interesting to see if they're good enough. Do you mean the kind that hang from the ear, sort of thing, or the - The k- kind that guys that like to look like they're really cool at airports wear. Yeah. Yeah. Yeah, but I don't know that one. So what's that mean? This - it's - it's a cell phone jack. Oh, yeah, you do. You can't miss it. You cannot miss it. They're the guys going around - they're probably talking to nobody, but Is this - just this little - Oh they - they have little headsets? Or what do they - Yep. Th- d- Yes. They wear the heads- Yep. E- Well, there you go. Yes. That's - that - it looks sort of like that. Oh you've - you've got one. Except that one's even bigger than most of the ones I've seen. Most of the ones are just a little boom. So where - where does - how do you wear that? What does that - You - you put it over the corner of your ear. Yeah, over the corner of the ear - I can't - I'm having competing microphone - headphone problems here. There's another kind which s- goes in your ear and it hangs down. There's a little b- bulb Huh. Yep. It just ha- just hangs down there. Yeah. Yeah, that I've seen. Yeah. that hangs - has the mike on it. O_K, so whatever people sort of This one. wear to use l- Yeah, something like tha- with their cell phones, Mm-hmm. Yeah. it'd be really great, I think, if we can argue t- that - It's an interesting idea. people like that. The acoustics - the acoustics on this are really s- m- better than when I talk into it directly. Yeah. The - the biggest issue is the stupid cable But - what - Are they? Why would it be cool? that Sony - The connector that Sony has, that, uh - Oh, O_K. So it's - they're not compatible? It's non-standard, and so you can't just plug something in. You need to get it wired. Oh, I'm sorry. You're talking about that as a possible thing to plug into the Sonys. No? Well, um, just as an example for, Right. How standard are these? Works with this. I don't know. you know, the future, of the fact that maybe people will wear those microphones, or some people might, to meetings, you know. Well, some people will, yeah. Not a - I'm not saying there's not a far-field microphone, uh, application, but - If we have a choi- I've always wondered how well they would work. It is an interesting idea. If we could build an adaptor that went from - cuz I think mine looks like this too. If we had this style plug that went to that, Right. then we'd just have adaptors and people would plug them in. You could use - they could use their own. Oh, right. So they could use the - I wonder if you could do an adaptor. I don't know if you can or not. Anyway, it's just an idea. Um - Yeah. An adaptor might be a better idea than do- redoing the wiring all the time. Mmm. Well, I mean, another s- An- another side to that is that, um, in principle, uh, if it all comes together, supposedly we're gonna get two um, I_RAM boards. Right. Uh, little later on in the year. And so, we're gonna wanna be thinking in terms of some wireless connection Mm-hmm. to it. Um, so we have, you know, some kind of little something, and, uh, we have a wireless connection, this board, that's doing the computation for the f- the, uh - the recognition. And so, you know, we wanna be able to show off to somebody, you know, we have a base station somewhere, and you're wandering around, uh, hands free, and - Yep. I don't understand. So - so there's a mike that plugs in - You're saying there's a mike that plugs in to the I_RAM board? or - No, no, what I'm saying is that we're gonna have to put together some kind of a little superstructure, d- some sort of demo box that it - that doesn't Demo. do a whole lot, but it somehow connects to - uh, remotely to this I_RAM board which will be doing the real computation. Oh, I see. In other words, we're - we're not gonna attempt to build, uh, it - do the real work of building a portable device that's little, that is doing all the computation we wanna do for the recognition and the far-field mike, uh, compensation and all that stuff. So we're gonna do all that on this powerful board, and, uh, have some wireless connection to it. And then the question is what is - what does it look like, does it look like something that you - you hold like this, or do we make a big deal out of it that you hold it arm - at arm's length th- so that you can see some display, or, uh, do we have one of these - You know. I don't know. So, it th- these are all possibilities. Yep. I mean, what I was envisioning was a P_D_A with a cellular link in it. You know, with one of the r- short-range wireless. And let th- Just do - capture the audio on the P_D_A and send the audio over the wireless net. Yeah. So, is- is Bluetooth the sort of thing that'd be, uh - But. Uh, Bluetooth is shorter range, but you could use Bluetooth. Wh- whatever the network guys already have hooked up for this floor. I don't remember what it's called. Hmm. O_K, then uh, another issue on file reorganization is, uh, making data available to people outside ICSI. So, specifically, the U_ W folks have been wanting to get access to it. So I think the right thing to do for that, is figure out how to do C_V_S without uh, compromising security. Some pass- S_S_H tunneled C_V_S, Well, I- I was - and then give them access. I was reading through some of the C_V_S documentation and you can just substitute C_V_S for your R_S_ H command. Yes, except that the command has to not take a password. Hmm. Because it- it's non-interactive. And so the only way to do that is with, uh - S_hosts , which is insecure. Hmm. So we still have to sort of look at it a little bit. I mean, I'm sure it's been solved, and I just haven't found the solution yet. But I am sure that people have done it because it's gonna be a problem a lot of people have. But, um - Who - Well, we're s- talking about a fairly limited group, are - are - is it a small enough group that they would - could just have accounts? Does that solve anything? Uh, n- n- not really, because then we have the coordination issue. I mean, so one thing they could do - I mean, I guess that's true. They could simply log in to ICSI Yeah. Well they're gonna have to have accounts - and do everything locally. If they use S_S_H they have to have an account anyways. Yeah, except they could share one. Well, but that's not - The, uh - the S_S_H accounts, and the user accounts don't have to be similar. So. I mean, there are lots of ways of doing it, but w- but if we can't figure out a remote way of doing it, just letting them log in to ICSI might be O_K. Mm-hmm. So the intention was to put everything except the audio files themselves under revision control. And the audio files, it's not worth doing cuz they're too big. You're never gonna be copying them around and making working copies of them. Right. And then the other issue Jane and I spoke of briefly is just general permission issues. That right now, lots of people create the files and then we have group "Meeting Recorder". But that means anyone in the Meeting Recorder group can overwrite the files. And so that's a pretty coarse level of granularity. So we might wanna think about doing a Meeting Recorder user - owner for those files, and then doing group slightly differently. But - Can you just do, like, S_U_X or something to modify - I think what we have now - I mean, if you just have one user that - Another option is to m- make a non- make it owned by a not real user, and then root will be able to do it. But root is pretty tightly controlled here. I had another thought. Mm-hmm. So th- I don't think that's a good solution. Would - w- would it be possi- There are scripts that uh, run, and - and when it's finished compressing, would it be possible for the script to change the permissions on the file to be more strict? Have them - Yes, but if you do that, then someone who isn't the owner can't unchange it. But after the meeting's done, would they want to? Yeah, as I said, it's just a question of, uh, do you wanna have to track down root if you have to make any change. But it doesn't have to be root. It could be some other user that we all can S_U_X to. That's true. You have security issues with that. So. Yeah, that's true. I mean - That's what we do with doctor-speech though, right? No, with doctor-speech, it's just a group. So - there're lots of different owners of i- i- of the files, but the group is always doctor-speech. Or real, usually. But there is a user doctor-speech. Right? I mean, I - I S_U_X to doctor-speech w- when I need to make a - Nnn- Oh, do you? Oh. So tha- that's what I meant, it - just some a directory on, uh, one of the - user like "meeting-root" or something. Oh, O_K. I don't know. I mean, I'm not too worried, because the group's pretty small. But - Yeah. Also it seems like once it's been stored that night, it's - it's possible to recover from backups, so that the main risk is avoiding it being accidentally deleted during the day that it was recorded. Well, and just - You, then, also just have the general problem of permission on the other files, that, do you want people checking out a transcript file when they find an error, and correcting it, or do they wa- do you want it to go through you? Yeah. Yeah. And I'd prefer it to go through me. O_K, but that means that you're gonna have to be available for people, all the time, who say "there's an error here", rather than just - them just going in and correcting it. Well I p- At this point - Yeah. Well, with C_V_S, though, you don't actually have to do it that way. Yeah. Well, but - You have a choice. Either you let people do it themselves, or you don't. Yeah. What I'm saying is that it can be the case that people can do it themselves and it can be reviewed. w- w- Well- Right? Right. I mean, it's not like it's one or the other. We can do both. Yeah, you can - Right, so you could get email. Send email to Jane that someone has done a Right. I - I wanna also say that, you know, this is a - revision. uh, uh, My - my view on this is that once it's in a final version and I'm working toward all these being final versions, then it can go into this next ro- this next realm. But it - it's sort of counter-productive to have changes made which I would be making if I'd just gotten that far in the file. You know, it's - it's, um. Well, so, you don't have to release it to the world until you're ready to. Yeah. That's right. So the question is s- w- And then after that, then this would probably work. Yeah. So you would keep it in a separate file structure until you were ready to put it in the repository. Good, good. I mean - Or just keep the lock on it. Is C_V_S, like - Yeah, just keep the lock on. I mean, just check it out - check it out and don't let anyone check it out - Until you're - Yep. No, it does not - the C_V_S doesn't work that way. Oh, it - Oh, it doesn't. Oh, O_K. There's no - there's no lock when you check something out, not like R_C_S. In other words, everybody can check out anything. Oh. The C_ stands for "concurrent". So why not use R_C_S at that point? Just at that stage? Cuz it's not remote. It's not remote. The nice thing about C_V_S is - is, you can be on a different machine. But at that point - we only want, probably, Jane to be in control anyway, and she - I mean, it's just an i- No, because U_W wants to have access to these files. That's the whole point. But not to modify them, right? Just to read them. Well, if they k- en- if they record a meeting, and they create a "key" file, and they find an error in their "key" file, they shouldn't have to tell us about that so that they can create - correct an error in one of their files. Well, the " key " file is - So that's why I wanna do it - I don't know, uh - I mean - ke- " key " file's different. "Key" fi- Different than what? Well, "key" file has the extra attributes that if you notice that there are lots of spikes on channel so-and-so, it's useful as a repository for putting that there. And also if you notice that there was an extra speaker, and didn't get mentioned in the "key" file, but they were only there for a second, but it's t- useful to mention that they were there. So I th- I think that the "key" file is usefu- for notations related to the meeting, Well - B- But, I mean, the meetings are gonna be the same way. So U_W records a meeting, we transcribe it. but it's different from the transcript. Yeah. A few months later they listen to it and they say "Ope! They got that acronym wrong." Yeah. Why - why sh- Then they should send that through - Well - It's sort of not really a question about permissions, but more of procedure. Either those all go through Jane, or through someone, Mm-hmm. or they all don't. But it's sort of - And if they all do, then there isn't a problem, right? Cuz once they give us the data, it's ours, and if they wanna make changes - I mean, ours to sort of transcribe and annotate. And if they wanna make changes, they can do that. The - the p- the point here is that it would be nice t- for there to be one repository. Yeah. Yeah And if we don't let them modify their own data, they're not gonna store their data in our repository. Or if we make it inconvenient for them to change their data. So then if they do have acc- I - I'm just looking at it the other way around. O_K, so what if they have - What if they have accounts here, So - so let's say we k- and they use R_C_S, at that point, where you can l- really lock a file. I mean, I'm worried. If you can't lock a file, this - this to me sounds very scary. Hmm. Um - And if they have accounts here and they're modifying it - If they're ac- if they're so closely linked that they're actually modifying transcripts and "key" files, then they could do it by, you know, secure-shelling into ICSI, under R_C_S, at that point. Ye- well, I don't see a real difference between R_C_S and any other system. I mean, it's just mechanisms. The nice thing about C_V_S is that you can have multiple people modifying and if the changes don't overlap each other, they can just do it. Hmm. That's interesting. So the problem with R_C_S is that if two people Yeah. want to modify a file at the same time they can't, because the granularity of locking is at the file level. What- what does C_V_S do if they - if they do overlap? Uh, there're lots of different policies for dealing with it. So - Mm-hmm. But typically a person has to, Yeah. uh, mediate Oh. you know, which changes actually get - go through. Yeah, to decide which one is - Yeah. Oh. Oh, I- So, maybe there is a way in C_V_S to effectively lock something if you don't want people to make any changes? At the repository level, there certainly is. You can mark a file as - as "you can't check this out". Uh-huh. Oh. So - C_V_S is pretty good. So then you could use C_V_S and, you know, just f- have remote access, but then it's up to whoever O_K. Yep. is, sort of, responsible for that level of transcription to decide how and when to put these locks, or - Well, it's - it's just like source code, right? So that, when you're developing it and it's really rough, you don't put it in the repository. Yeah. You wait to the point when you're ready to release it, and then you put it in the repository and if other people wanna - Yeah, so the w- Mm-hmm. In- The - Maybe the model - It's a little bit different in C_V_S. When you check something out, you actually create your own directory with copies of everything in it, and that's what you work on. Copy. So, until you're ready to check it in, nobody sees anything you've done. And then when you check it in, it puts, sort of, you know - every change that you've done is - goes into the central - Right. So if that's at an- Hmm. I guess it depends on Jane's - Uh, you know, if that's - if that model works for the transcripts, then that's fine. But if that pr- @@ um, allows someone to come in and modify while you're modifying, and they turn out to be changes that, you know, would have been better to wait until your version came out, then that's really up to you, not up to the software, so. Well, I also think, um, that the type of change ch- ch- varies an- I think it's a very good point. Because what if the - uh, what if one set of changes one person was making was with reference to some typographical convention. Mm-hmm. And the other person n- unkn- n- quite unknowingly, was changing it the other way around. Then you could end up with a real mess, in terms of, like, consistency of conventions across a file, there is an argument to be made for having a single editor editing at that level. Well yeah, there would - there would have to be somebody to enforce the consistency. Yeah. But I like b- I like what Liz is saying about - That's right. And I also like - I l- also like Liz's uh, k- um, phraseology of "at this level of" - I don't remember how you put it, but this - "at this level of transcription" or whatever. I mean it's like there are different layers of whatever and different types of conventions, and if it were an acronym, then I would - I think that it would make sense f- for them to have input into that level of things, but I wouldn't want them to be changing - I'm just - Imagine that we reverse this and they were keeping the repository and we were d- collecting meetings and sending them to them. Yeah. Mm-hmm. If I found an error, and I sent them email and said "here's this error", and I didn't get a response for a couple weeks, I would stop sending them the - my data and I would start collecting it here and we would end up with two corpora. Well, yeah. And I don't want that to happen. So we have to make it convenient for them to make changes that they wanna make. I agree. Well - Well, I see now, we're k- back into the same situation as when we were, uh, talking about a- allowing people to edit the transcripts before they become public. So there's a question of how easy or how hard to modify, and how that Mm-hmm. impacts the amount of changing that's done. And I - I think that it's, um, I - I have received some suggestions for change, which I've made. And um, I think with an acronym, that's a clear case where it would be an easy change to incorporate. But it would worry me to have people, um, having free reign to change these things, because of the need to have consistent conventions. And people who are only just visiting the data might not have a sense of the larger system. Yeah, that's what I would be worried about too. Well, I think - yeah, I think there are probably ways that you can say that, you know, only certain users are allowed to change Yeah. things. Mm-hmm. Uh, so we can probably restrict it to be - You know, f- casual users who are just browsing can just r- get it read-only, can't check in to the archive. Mm-hmm. Whereas we could say "other people", you know, "these people are allowed to actually submit changes". Right. I mean, I - I think one thing that's probably everybody would agree on is that there will never be a point where you know, we can say that for sure there's no problems with any of the transcripts. I mean there's, uh - f- You know, something's going to get overlooked at some point, or somebody's gonna wanna make a change somehow. So we need to have some mechanism for, um, handling changes in the future. I agr- I agree but I have to also add one other thing which i- before I forget it, which is Adam's point about when you were reviewing the I_B_M transcripts. u- This was mentioned at the beginning of the meeting. You said that there were some places where it wasn't what y- was written already, but you could see how they might have thought that's what it said. So there - it's not just a right and wrong thing. Sometimes there tru- there are absolutely stretches where there is no - no - k- if you were at the meeting and you could ask the person, Yeah. Exactly. Mm-hmm. maybe, m- you know, they could - they could've told you, but five minutes later they might not have been able to tell you, Right. even if they were listening to it. So, sometimes it isn't right and wrong, sometimes it's multiple interpretations, and - Yeah. Good point. Yeah. Now - But of course, this does raise the possibility of allowing, um, an extra annotation for, you know, alternative interpretations of a stretch. Mm-hmm. But I just wanted to say it isn't simply right or wrong. Yeah, that's a good point. Or overlooked or not overlooked. Sometimes they're truly ambiguous, and Mm-hmm. um, uh, um, nice to have a notation for that, but. Yeah. Gotta go. Yeah. I mean, I like that idea because even a casual user can always send email Mm-hmm. to whoever's in charge, and say, you know, "we'd like these changes" and, Well - you know, hopefully we'll give them a response. And if they really do it a lot, and they say I just - I don't s- I don't see that this is e- Yep. "we're a casual user but we want a chance to change the transcripts " then we can face that if it happens, but Can you write "unread" on that? Yeah, thank you. I don't really see the people at U_W that I know of right now making huge amoun- investing huge amounts of time in changing transcripts. But I could be wrong. Probably not transcripts, but it would be nice to have one mechanism for all the files. And they're certainly gonna wanna change tools. Yeah. And so - you know, it's the same as source code. You - you release source code at some point. And, either you have to do everything or you share responsibility. Yeah I mean a lot of the open source pro- projects face this same problem. Right. You know, p- anybody can check out the code, but not everybody can check it in. Yeah. Right. And there's some happy medium and we don't know what that is yet until we get feedback from people, but what if it's O_K to just handle it with sort of a person in charge of the philosophy behind the changes, and some people with permissions, maybe by request, to make changes, That we don't just give people permissions if they're not gonna make changes. Because I've overwritten a - a file by mistake, not wanting to have done that cuz I didn't think I had permission, when I did. Mm-hmm. Um, and then just Mm-hmm. seeing if that is enough to - to handle the transcript changes. I'm just worried about letting everybody go in and make changes, cuz it's real easy when you're trying to, I don't know, run alignments and there's a word you wanna fix, to go in and do that and then mess up other things, if you don't know, you know, the overall philosophy behind the - the conventions. Mm-hmm. Right. And als- it also- Mm-hmm. Then - There is also the possibility - Remember, in C_V_S, when we talk about making changes, You can back everything out. there's sort of two types that can be made. To your local, or to the global. Mm-hmm. And so people can make changes to their local and if they screw those up, Right. Exactly. that's, you know, in their - that's only on them. It's w- it's the checking in part that r- we really care about, and Mm-hmm. that we can control with, you know, who can check things in and stuff. So we can - I think we can do this. Or just, yeah, start by make it a really tight control and then as people really need the control you can Mm-hmm. Mm-hmm. Yeah. They can still check it out - ascertain whether or not to - Yeah, cuz they can still check out full copies and make whatever changes they want to their local copies. Right. And it won't affect the - the g- the global, official, ones at all. Right. Right, but what I want to avoid is ending up with two corpora. Yeah. Yes. And so if you make it too hard for them to check back in, they'll check it out once, Definitely. Yep. they'll make all the changes, they'll never tell us about the changes, Mm-hmm. and we'll get two different corpora. Mm-hmm. Right? I think we're already gonna - I mean there's already some chance that different annotations, different places, you know - But you can control that by knowing you're making two corpora or knowing that you're adding um, annotations on one version and you don't have the latest corrections maybe at that point. And then you finish the project and you realize that there were corrections made on your originals and then you have to merge them. And the b- the thing that makes it ea- O_K to do that, is knowing where the synch time boundaries are. Cuz you can automatically pretty much merge things if you've only got twenty words or so in an utterance. It's when you get the whole meeting and the synch times ch- have changed, or you can't correspond to a previous version with synch times, that you get in trouble. So. Was there anything else on your list? Nope. Shall we do some digits? Uh, Transcript L_ one one six. Now, wait, are - are we doing them simultaneously or one at a time? Alright. I wasn't planning on doing it simultaneously. Lee , you liked that l- huh? Transcript L_ one one six. eight two six, four four five, one nine nine one three one one, eight seven three six, seven one one seven eight, two eight three, eight zero, three eight two, nine two, seven five eight, four three, zero nine nine, seven zero nine four, four eight four, four seven four seven six, four one, eight eight, nine nine, eight seven three two, four six, four three, one nine, two eight four seven six two, four O_ four five, nine six five one. Transcript L_ one one seven. seven five eight, five one three, O_ seven one four two six nine four, five six two three, seven O_ four four seven, five one three, seven four, eight O_ one, five six six two seven, five, three seven six eight O_ six, nine seven six, one four two one three nine, eight five, two three one O_ eight eight four four, four, two nine five four six one, one O_ three, two two four. Transcript L_ dash one two seven. two eight two, two seven six, eight six four eight eight three, two zero six, two six two eight seven, five six, seven six, one seven, zero zero seven eight three zero, seven five two seven, five three eight three five five, seven three, zero eight, five nine, five seven one, four eight two, eight nine, six six seven, zero two nine three six, nine two one five, zero five eight zero nine seven six, six O_ three, nine seven six. Transcript L_ one two nine. zero nine, zero zero, five three, nine two, five five six, zero three seven, seven zero, nine eight eight, nine zero seven five, nine six nine, one one eight two six eight seven, two five, one one four four O_ four four, five five two, seven five O_ five five, seven five three, O_ O_, six seven nine one seven eight four, three, two seven eight six three seven five, one, two eight four. Transcript L_ one two eight. eight, four three two, three six, one O_ seven, six zero four six, six nine, six five O_ nine six two, eight seven, eight one, seven seven, O_ two nine, seven nine two, one zero, zero five five, one three zero, four seven, nine, six one five two zero seven two, six zero, six zero, six seven, four two three, five two four, zero six, six O_ two, eight three eight two, two seven, one nine nine six. Oop! Ouch! Ooo!