10 examples of 'how to make speech recognition in python faster' in Python

47
def main():
48
    """Run all checks and print status."""
49
    if not os.path.exists(CREDENTIALS_PATH):
50
        print(
51
            """Please follow the Custom Voice User Interface instructions on the AIY website
52
to download credentials:
53
https://aiyprojects.withgoogle.com/voice-v1/#makers-guide-3-custom-voice-user-interface
54
and save them to""", CREDENTIALS_PATH)
55
        return
56
57
    if not check_credentials_valid():
58
        print(
59
            CREDENTIALS_PATH, """is not valid, please check that you have downloaded JSON
60
service credentials.""")
61
        return
62
63
    if not check_speech_reco():
64
        print('Failed to test the Cloud Speech API. Please see error above.')
65
        return
66
67
    print("Everything is set up to use the Google Cloud.")

238
def main():
239
    """start bidirectional streaming from microphone input to speech API"""
240
241
    client = speech.SpeechClient()
242
    config = speech.types.RecognitionConfig(
243
        encoding=speech.enums.RecognitionConfig.AudioEncoding.LINEAR16,
244
        sample_rate_hertz=SAMPLE_RATE,
245
        language_code='en-US',
246
        max_alternatives=1)
247
    streaming_config = speech.types.StreamingRecognitionConfig(
248
        config=config,
249
        interim_results=True)
250
251
    mic_manager = ResumableMicrophoneStream(SAMPLE_RATE, CHUNK_SIZE)
252
    print(mic_manager.chunk_size)
253
    sys.stdout.write(YELLOW)
254
    sys.stdout.write('\nListening, say "Quit" or "Exit" to stop.\n\n')
255
    sys.stdout.write('End (ms)       Transcript Results/Status\n')
256
    sys.stdout.write('=====================================================\n')
257
258
    with mic_manager as stream:
259
260
        while not stream.closed:
261
            sys.stdout.write(YELLOW)
262
            sys.stdout.write('\n' + str(
263
                STREAMING_LIMIT * stream.restart_counter) + ': NEW REQUEST\n')
264
265
            stream.audio_input = []
266
            audio_generator = stream.generator()
267
268
            requests = (speech.types.StreamingRecognizeRequest(
269
                audio_content=content)for content in audio_generator)
270
271
            responses = client.streaming_recognize(streaming_config,
272
                                                   requests)
273
274
            # Now, put the transcription responses to use.
275
            listen_print_loop(responses, stream)
276
277
            if stream.result_end_time > 0:
278
                stream.final_request_end_time = stream.is_final_end_time
279
            stream.result_end_time = 0
280
            stream.last_audio_input = []
281
            stream.last_audio_input = stream.audio_input
282
            stream.audio_input = []
283
            stream.restart_counter = stream.restart_counter + 1
284
285
            if not stream.last_transcript_was_final:
286
                sys.stdout.write('\n')
287
            stream.new_stream = True

26
def test_google_french(self):
27
    r = sr.Recognizer()
28
    with sr.AudioFile(self.AUDIO_FILE_FR) as source: audio = r.record(source)
29
    self.assertEqual(r.recognize_google(audio, language="fr-FR"), u"et c'est la dictée numéro 1")

31
def test_google_chinese(self):
32
    r = sr.Recognizer()
33
    with sr.AudioFile(self.AUDIO_FILE_ZH) as source: audio = r.record(source)
34
    self.assertEqual(r.recognize_google(audio, language="zh-CN"), u"砸自己的脚")

85
def speech_synthesis_with_voice():
86
    """performs speech synthesis to the default speaker with specified voice"""
87
    # Creates an instance of a speech config with specified subscription key and service region.
88
    speech_config = speechsdk.SpeechConfig(subscription=speech_key, region=service_region)
89
    # Sets the synthesis voice name.
90
    # The full list of supported voices can be found here:
91
    # https://docs.microsoft.com/azure/cognitive-services/speech-service/language-support#text-to-speech
92
    voice = "Microsoft Server Speech Text to Speech Voice (en-US, BenjaminRUS)"
93
    speech_config.speech_synthesis_voice_name = voice
94
    # Creates a speech synthesizer for the specified voice,
95
    # using the default speaker as audio output.
96
    speech_synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_config)
97
98
    # Receives a text from console input and synthesizes it to speaker.
99
    while True:
100
        print("Enter some text that you want to speak, Ctrl-Z to exit")
101
        try:
102
            text = input()
103
        except EOFError:
104
            break
105
        result = speech_synthesizer.speak_text_async(text).get()
106
        # Check result
107
        if result.reason == speechsdk.ResultReason.SynthesizingAudioCompleted:
108
            print("Speech synthesized to speaker for text [{}] with voice [{}]".format(text, voice))
109
        elif result.reason == speechsdk.ResultReason.Canceled:
110
            cancellation_details = result.cancellation_details
111
            print("Speech synthesis canceled: {}".format(cancellation_details.reason))
112
            if cancellation_details.reason == speechsdk.CancellationReason.Error:
113
                print("Error details: {}".format(cancellation_details.error_details))

111
def transcribe_file_with_auto_punctuation():
112
    """Transcribe the given audio file with auto punctuation enabled."""
113
    # [START speech_transcribe_auto_punctuation_beta]
114
    from google.cloud import speech_v1p1beta1 as speech
115
    client = speech.SpeechClient()
116
117
    speech_file = 'resources/commercial_mono.wav'
118
119
    with io.open(speech_file, 'rb') as audio_file:
120
        content = audio_file.read()
121
122
    audio = speech.types.RecognitionAudio(content=content)
123
    config = speech.types.RecognitionConfig(
124
        encoding=speech.enums.RecognitionConfig.AudioEncoding.LINEAR16,
125
        sample_rate_hertz=8000,
126
        language_code='en-US',
127
        # Enable automatic punctuation
128
        enable_automatic_punctuation=True)
129
130
    response = client.recognize(config, audio)
131
132
    for i, result in enumerate(response.results):
133
        alternative = result.alternatives[0]
134
        print('-' * 20)
135
        print('First alternative of result {}'.format(i))
136
        print('Transcript: {}'.format(alternative.transcript))

107
def main():
108
    # test example address file
109
    plaintext = 'resources/example.txt'
110
    ssml_text = text_to_ssml(plaintext)
111
    ssml_to_audio(ssml_text, 'resources/example.mp3')

54
def speech_synthesis_with_language():
55
    """performs speech synthesis to the default speaker with specified spoken language"""
56
    # Creates an instance of a speech config with specified subscription key and service region.
57
    speech_config = speechsdk.SpeechConfig(subscription=speech_key, region=service_region)
58
    # Sets the synthesis language.
59
    # The full list of supported languages can be found here:
60
    # https://docs.microsoft.com/azure/cognitive-services/speech-service/language-support#text-to-speech
61
    language = "de-DE";
62
    speech_config.speech_synthesis_language = language
63
    # Creates a speech synthesizer for the specified language,
64
    # using the default speaker as audio output.
65
    speech_synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_config)
66
67
    # Receives a text from console input and synthesizes it to speaker.
68
    while True:
69
        print("Enter some text that you want to speak, Ctrl-Z to exit")
70
        try:
71
            text = input()
72
        except EOFError:
73
            break
74
        result = speech_synthesizer.speak_text_async(text).get()
75
        # Check result
76
        if result.reason == speechsdk.ResultReason.SynthesizingAudioCompleted:
77
            print("Speech synthesized to speaker for text [{}] with language [{}]".format(text, language))
78
        elif result.reason == speechsdk.ResultReason.Canceled:
79
            cancellation_details = result.cancellation_details
80
            print("Speech synthesis canceled: {}".format(cancellation_details.reason))
81
            if cancellation_details.reason == speechsdk.CancellationReason.Error:
82
                print("Error details: {}".format(cancellation_details.error_details))

68
def transcribe(speech_file):
69
    """Transcribe the given audio file."""
70
71
    speech_client = speech.Client()
72
73
    with io.open(speech_file, 'rb') as audio_file:
74
        content = audio_file.read()
75
        audio_sample = speech_client.sample(
76
            content=content,
77
            source_uri=None,
78
            encoding='OGG_OPUS',
79
            sample_rate_hertz=16000)
80
81
    alternatives = audio_sample.recognize('es-ES')
82
    for alternative in alternatives:
83
        return '{}'.format(alternative.transcript)

80
def speak_google_tts(text):
81
    """ This method implements Text to Speech using the Google Translate TTS.
82
    It uses Google Speech Python Package.
83
    :param text: Text which is needed to be spoken
84
    :return: None
85
    """
86
    with tempfile.TemporaryDirectory() as tmpdirname:
87
        fd, mpiii = tempfile.mkstemp(suffix='.mp3', dir=tmpdirname)
88
        Speech(text=text, lang=susi_config["language"]).save(mpiii)
89
        player.say(mpiii)