File size: 4,687 Bytes
a7c4935
 
 
 
 
 
 
 
 
177a062
a7c4935
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
177a062
a7c4935
 
 
 
 
177a062
 
a7c4935
 
 
 
 
177a062
 
a7c4935
 
 
 
 
 
177a062
a7c4935
 
 
 
177a062
a7c4935
177a062
a7c4935
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
177a062
 
a7c4935
177a062
a7c4935
 
 
 
177a062
a7c4935
 
 
 
177a062
 
 
 
 
 
 
 
 
 
 
a7c4935
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
from urllib.parse import urlparse, urlencode
from wsgiref.handlers import format_date_time
from datetime import datetime
import hmac
import hashlib
import base64
import numpy as np
import json
import websockets.client
from loguru import logger
from websockets.exceptions import ConnectionClosedError
import time

STATUS_FIRST_FRAME = 0
STATUS_CONTINUE_FRAME = 1
STATUS_LAST_FRAME = 2

'''
XFYun's IAT Client, which is used to convert speech to text.
'''
class IATClient:
    def __init__(
        self,
        app_id: str,
        api_key: str,
        api_secret: str,
        endpoint="wss://ws-api.xfyun.cn/v2/iat",
    ) -> None:
        self.app_id = app_id
        self.api_key = api_key
        self.api_secret = api_secret
        self.endpoint = endpoint
        self.common_args = {"app_id": self.app_id}
        self.business_args = {
            "domain": "iat",
            "language": "zh_cn",
            "accent": "mandarin",
            "vinfo": 1,
            "vad_eos": 10000,
        }

    # To convert ndarray audio data to PCM-style bytes
    # Gradio Audio Module returns a tuple of (sampling_rate, np.ndarray)
    # And the np.ndarray is the audio data, which is range from -32768 to 32767 matching PCM range.
    def encode_pcm(self, source: np.ndarray):
        return source.astype(np.int16).tobytes()

    def create_url(self):
        parse_result = urlparse(self.endpoint)
        host = parse_result.hostname
        # RFC1123 Timestamp
        date = format_date_time(time.mktime(datetime.now().timetuple()))
        path = parse_result.path

        sign_raw_str = f"host: {host}\ndate: {date}\nGET {path} HTTP/1.1"
        logger.debug(f"Sign raw string: {sign_raw_str}")
        sign_sha = hmac.new(
            self.api_secret.encode("utf-8"),
            sign_raw_str.encode("utf-8"),
            digestmod=hashlib.sha256,
        ).digest()
        sign_sha = base64.b64encode(sign_sha).decode("utf-8")
        auth_raw_str = 'api_key="%s", algorithm="%s", headers="%s", signature="%s"' % (
            self.api_key,
            "hmac-sha256",
            "host date request-line",
            sign_sha,
        )
        logger.debug(f"Authorization: {auth_raw_str}")
        auth = base64.b64encode(auth_raw_str.encode("utf-8")).decode("utf-8")
        params = {
            "authorization": auth,
            "date": date,
            "host": host,
        }
        url = f"{self.endpoint}?{urlencode(params)}"
        logger.debug(f"URL: {url}")
        return url

    def prepare_data(self, audio: bytes, chunk_size=1280, sampling_rate=16000):
        status = STATUS_FIRST_FRAME
        logger.debug(f"Total audio length: {len(audio)}")
        for i in range(0, len(audio), chunk_size):
            logger.debug(f"Processing chunk {i} to {i + chunk_size}")
            chunk = audio[i : i + chunk_size]
            if i + chunk_size >= len(audio):
                status = STATUS_LAST_FRAME
            data = {
                "status": status,
                "format": f"audio/L16;rate={sampling_rate}",
                "audio": base64.b64encode(chunk).decode("utf-8"),
                "encoding": "raw",
            }
            payload = {"data": data}
            if status == STATUS_FIRST_FRAME:
                payload["common"] = self.common_args
                payload["business"] = self.business_args
            yield payload
            status = STATUS_CONTINUE_FRAME

    async def dictate(self, audio: tuple[int, np.ndarray], interval=0.04):
        logger.debug(f"Generate URL")
        url = self.create_url()
        logger.debug("Encoding audio to PCM")
        sampling_rate, source = audio
        pcm = self.encode_pcm(source)
        async with websockets.client.connect(url) as ws:
            for payload in self.prepare_data(pcm, sampling_rate=sampling_rate):
                logger.debug('Sending payload')
                await ws.send(json.dumps(payload))
                time.sleep(interval)
            try:
                async for message in ws:
                    data: dict = json.loads(message)
                    logger.debug(f"Received data: {data}")
                    if not 'data' in data.keys():
                        yield ''
                        break
                    is_end = data["data"]["status"] == STATUS_LAST_FRAME
                    ws_list = data["data"]["result"]["ws"]
                    text = ''.join([cw["w"] for cw in sum([ws["cw"] for ws in ws_list], [])])
                    yield text
                    if is_end:
                        break
            except ConnectionClosedError as e:
                print(f"Connection closed: {e.code} {e.reason}")