py6WFXP2

1import xml.etree.ElementTree as XmlElementTree
2import httplib2
3import uuid
4from config import ***
5
6***_HOST = '***'
7***_PATH = '/***_xml'
8CHUNK_SIZE = 1024 ** 2
9
10
11def speech_to_text(filename=None, bytes=None, request_id=uuid.uuid4().hex, topic='notes', lang='ru-RU',
12            key=***_API_KEY):
13        '''Translate speech from file to text
14
15        Keyword arguments:
16        filename -- audio file with speech
17        bytes -- speech audio in bytes in pcm 16000 GHz
18        request_id -- uuid for connection to converting api
19        topic -- speech topic
20        lang -- speech language
21
22        Raised exceptions:
23        Exception -- raised when neither file name nor bytes provided
24        SpeechException -- raised when error in audio to text converting
25        '''
26        if filename:
27            with open(filename, 'br') as file:
28                bytes = file.read()
29        if not bytes:
30            raise Exception('Neither file name nor bytes provided.')
31        bytes = convert_to_pcm16b16000r(in_bytes=bytes)
32        url = ***_PATH + '?uuid=%s&key=%s&topic=%s&lang=%s' % (
33            request_id,
34            key,
35            topic,
36            lang
37        )
38        chunks = read_chunks(CHUNK_SIZE, bytes) # splitting bytes to processing chunks
39
40        # openning connection to api for sending chunks to converting api
41        connection = httplib2.HTTPConnectionWithTimeout(***_HOST)
42        connection.connect()
43        connection.putrequest('POST', url)
44        connection.putheader('Transfer-Encoding', 'chunked')
45        connection.putheader('Content-Type', 'audio/x-pcm;bit=16;rate=16000')
46        connection.endheaders()
47
48        # sending chunks to api
49        for chunk in chunks:
50            connection.send(('%s\r\n' % hex(len(chunk))[2:]).encode())
51            connection.send(chunk)
52            connection.send('\r\n'.encode())
53        connection.send('0\r\n\r\n'.encode())
54
55        response = connection.getresponse()
56        if response.code == 200:
57            response_text = response.read()
58            xml = XmlElementTree.fromstring(response_text)
59
60            if int(xml.attrib['success']) == 1:
61                max_confidence = - float("inf")
62                text = ''
63
64                # searching text with largest confidence
65                for child in xml:
66                    if float(child.attrib['confidence']) > max_confidence:
67                        text = child.text
68                        max_confidence = float(child.attrib['confidence'])
69
70                # if no text in response raise SpeechException
71                if max_confidence != - float("inf"):
72                    return text
73                else:
74                    raise SpeechException('No text found.\n\nResponse:\n%s' % (response_text))
75            else:
76                raise SpeechException('No text found.\n\nResponse:\n%s' % (response_text))
77        else:
78            # raise SpeechException because of api internal error
79            raise SpeechException('Unknown error.\nCode: %s\n\n%s' % (response.code, response.read()))
80
81сlass SpeechException(Exception):
82    pass
83
84
85