def Translate(targetLanguage):
translation = ''
# Translate speech
audio_config = speech_sdk.AudioConfig(use_default_microphone=True)
translator = speech_sdk.translation.TranslationRecognizer(
translation_config, audio_config)
print("Speak now...")
result = translator.recognize_once_async().get()
print('Translating "{}"'.format(result.text))
translation = result.translations[targetLanguage]
print(translation)
# For audio file
# Translate speech
audioFile = 'station.wav'
playsound(audioFile)
audio_config = speech_sdk.AudioConfig(filename=audioFile)
translator = speech_sdk.translation.TranslationRecognizer(
translation_config, audio_config)
print("Getting speech from file...")
result = translator.recognize_once_async().get()
print('Translating "{}"'.format(result.text))
translation = result.translations[targetLanguage]
print(translation)
# Synthesize translation
voices = {"fr": "fr-FR-Julie", "es": "es-ES-Laura", "hi": "hi-IN-Kalpana"}
speech_config.speech_synthesis_voice_name = voices.get(targetLanguage)
speech_synthesizer = speech_sdk.SpeechSynthesizer(speech_config)
speak = speech_synthesizer.speak_text_async(translation).get()
if speak.reason != speech_sdk.ResultReason.SynthesizingAudioCompleted:
print(speak.reason)
def transcribe_audio_file_path(self, audio_file_path):
# For now supports wav, not mp3
# https://stackoverflow.com/questions/51614216/what-audio-formats-are-supported-by-azure-cognitive-services-speech-service-ss?rq=1
audio_config = speechsdk.AudioConfig(use_default_microphone=False,
filename=audio_file_path)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=speech_config, audio_config=audio_config)
transcript = ""
transcription_status = TranscriptionStatus.success
try:
result = speech_recognizer.recognize_once()
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
transcript = result.text
elif result.reason == speechsdk.ResultReason.NoMatch:
transcription_status = TranscriptionStatus.transcription_error
elif result.reason == speechsdk.ResultReason.Canceled:
transcription_status = TranscriptionStatus.unknown_error
except:
print("Unknown transcription error:", sys.exc_info())
transcription_status = TranscriptionStatus.unknown_error
return transcript, transcription_status
def from_file():
speech_config = speechsdk.SpeechConfig(
subscription="3785c3171b694795ad5e7ec2bc01ab7f",
region="canadacentral")
audio_input = speechsdk.AudioConfig(
filename=
'sound/audio_verification_challenge_35460110d06c257a9.9555557505-5083.wav'
)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
print("Recognized: {}".format(result.text))
return result.text
elif result.reason == speechsdk.ResultReason.NoMatch:
print("No speech could be recognized: {}".format(
result.no_match_details))
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
print("Speech Recognition canceled: {}".format(
cancellation_details.reason))
if cancellation_details.reason == speechsdk.CancellationReason.Error:
print("Error details: {}".format(
cancellation_details.error_details))
def TranscribeCommand():
command = ''
# Configure speech recognition
# For microphone
# audio_config = speech_sdk.AudioConfig(use_default_microphone=True)
# speech_recognizer = speech_sdk.SpeechRecognizer(speech_config, audio_config)
# print('Speak now...')
# Configure speech recognition
audioFile = 'time.wav'
playsound(audioFile)
audio_config = speech_sdk.AudioConfig(filename=audioFile)
speech_recognizer = speech_sdk.SpeechRecognizer(speech_config,
audio_config)
# Process speech input
speech = speech_recognizer.recognize_once_async().get()
if speech.reason == speech_sdk.ResultReason.RecognizedSpeech:
command = speech.text
print(command)
else:
print(speech.reason)
if speech.reason == speech_sdk.ResultReason.Canceled:
cancellation = speech.cancellation_details
print(cancellation.reason)
print(cancellation.error_details)
# Return the command
return command
def TranscribeCommand():
command = 'stop.'
# Configure speech recognition
audio_config = speech_sdk.AudioConfig(use_default_microphone=True)
source_language_config = speech_sdk.languageconfig.SourceLanguageConfig(
"es-ES")
speech_recognizer = speech_sdk.SpeechRecognizer(
speech_config=speech_config,
audio_config=audio_config,
source_language_config=source_language_config)
# Process speech input
print('Say "stop" to end...')
speech = speech_recognizer.recognize_once_async().get()
if speech.reason == speech_sdk.ResultReason.RecognizedSpeech:
command = speech.text
print(command)
print(Translate(command))
else:
print(speech.reason)
if speech.reason == speech_sdk.ResultReason.Canceled:
cancellation = speech.cancellation_details
print(cancellation.reason)
print(cancellation.error_details)
# Return the command
return Translate(command)
def process_recorded_data(file_q: Queue) -> None:
# load server configs
configs = {}
with open(config_file_path, 'r') as config_file:
c = yaml.safe_load(config_file)
configs['key'] = c['azure_subscription_key']
configs['endpoint'] = c['azure_endpoint']
configs['region'] = c['azure_region']
speech_config = speechsdk.SpeechConfig(region=configs['region'],
subscription=configs['key'])
source_language_config = speechsdk.languageconfig.SourceLanguageConfig(
"zh-CN", configs['endpoint'])
counter = Counter({"可以": 0, "吗": 0})
while True:
new_file_path = file_q.get()
# send it to microsoft
audio_input = speechsdk.AudioConfig(filename=new_file_path)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=speech_config,
source_language_config=source_language_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once()
if result.reason != speechsdk.ResultReason.RecognizedSpeech:
print(result)
print(result.text)
for k in counter:
if k in result.text:
counter[k] += 1
print(f'keyima counter {min(counter.values())}')
# remove the processed file
Path(new_file_path).unlink()
def audio_config_from_user_config(
user_config: helper.Read_Only_Dict) -> helper.Read_Only_Dict:
if user_config["input_file"] is None:
return speechsdk.AudioConfig(
use_default_microphone=True), None, None, None
else:
audio_stream_format = None
if not user_config["use_compressed_audio"]:
reader = wave.open(user_config["input_file"], mode=None)
audio_stream_format = speechsdk.audio.AudioStreamFormat(
samples_per_second=reader.getframerate(),
bits_per_sample=reader.getsampwidth() * 8,
channels=reader.getnchannels())
reader.close()
else:
audio_stream_format = speechsdk.audio.AudioStreamFormat(
compressed_stream_format=user_config["compressed_audio_format"]
)
callback = helper.BinaryFileReaderCallback(
filename=user_config["input_file"])
stream = speechsdk.audio.PullAudioInputStream(
pull_stream_callback=callback, stream_format=audio_stream_format)
# We return the BinaryFileReaderCallback, AudioStreamFormat, and PullAudioInputStream
# because we need to keep them in scope until they are actually used.
return helper.Read_Only_Dict({
"audio_config":
speechsdk.audio.AudioConfig(stream=stream),
"audio_stream_format":
audio_stream_format,
"pull_input_audio_stream_callback":
callback,
"pull_input_audio_stream":
stream,
})
def recognize(self, audio_path, settings, result_callback):
if not settings['key']:
result_callback("ERROR: No API key provided")
result_callback(self.END_VAL)
return
speech_config = speechsdk.SpeechConfig(subscription=settings['key'],
region=settings['region'])
audio_input = speechsdk.AudioConfig(filename=audio_path)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=speech_config,
audio_config=audio_input,
language=settings['language'])
def end_callback(evt):
nonlocal audio_input
nonlocal speech_recognizer
speech_recognizer.stop_continuous_recognition()
result_callback(self.END_VAL)
# fix thread not releasing audio file
speech_recognizer = None
audio_input = None
speech_recognizer.recognized.connect(
lambda evt: result_callback(evt.result.text))
speech_recognizer.session_stopped.connect(end_callback)
speech_recognizer.start_continuous_recognition()
def get_text(filepath):
speech_config = speechsdk.SpeechConfig(subscription=SUBSCRIPTION,
region=REGION)
speech_config.speech_recognition_language = 'ja-JP'
audio_input = speechsdk.AudioConfig(filename=filepath)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
done = False
text = ''
def stop_cb(evt):
print('CLOSING on {}'.format(evt))
speech_recognizer.stop_continuous_recognition()
nonlocal done
done = True
def split(evt):
st = re.search(r'\".+?\"', str(evt))
new_text = st.group(0).strip('"')
nonlocal text
text = text + '\n' + new_text
speech_recognizer.recognized.connect(split)
speech_recognizer.session_stopped.connect(stop_cb)
speech_recognizer.canceled.connect(stop_cb)
speech_recognizer.start_continuous_recognition()
while not done:
time.sleep(.5)
speech_recognizer.stop_continuous_recognition()
return text
def recognize(audio_filename):
audio_input = speechsdk.AudioConfig(filename=audio_filename)
# Creates a recognizer with the given settings
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
print("Recognizing first result...")
# Starts speech recognition, and returns after a single utterance is recognized. The end of a
# single utterance is determined by listening for silence at the end or until a maximum of 15
# seconds of audio is processed. The task returns the recognition text as result.
# Note: Since recognize_once() returns only a single utterance, it is suitable only for single
# shot recognition like command or query.
# For long-running multi-utterance recognition, use start_continuous_recognition() instead.
result = speech_recognizer.recognize_once()
# Checks result.
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
print("Recognized:")
ans = result.text
elif result.reason == speechsdk.ResultReason.NoMatch:
ans = "No speech could be recognized: {}".format(
result.no_match_details)
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
ans = "Speech Recognition canceled: {}".format(
cancellation_details.reason)
if cancellation_details.reason == speechsdk.CancellationReason.Error:
ans += "\nError details: {}".format(
cancellation_details.error_details)
return ans
def text_from_voice(file):
if file[-4:] != ".wav": # sanity check
return ({"msg": "file type error", "status": "ERROR"})
speech_config = speechsdk.SpeechConfig(
subscription=api_keys["microsoft-speech"]["key"],
region=api_keys["microsoft-speech"]["region"])
audio_input = speechsdk.AudioConfig(filename=file)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
data = {"msg": "", "status": "ERROR"}
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
data["msg"] = result.text
data["status"] = "OK"
elif result.reason == speechsdk.ResultReason.NoMatch:
data["msg"] = "No speech could be recognized: {}".format(
result.no_match_details)
data["status"] = "FAILED"
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
data["msg"] = "Speech Recognition canceled: {}".format(
cancellation_details.reason)
data["status"] = "CANCELED"
if cancellation_details.reason == speechsdk.CancellationReason.Error:
data["msg"] = ("Error details: {}".format(
cancellation_details.error_details))
data["status"] = "ERROR"
return data
def recognize_continuous(show: bool, file_name):
# use file for stt:
audio_input = speechsdk.AudioConfig(filename=f"data/{file_name}")
# set address of container
speech_config = speechsdk.SpeechConfig(host="ws://localhost:5000")
# request timing info
speech_config.request_word_level_timestamps()
# state of job, once finished == true
done = False
# instantiate speech_recognizer
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input)
def stop_cb(evt) -> None:
"""
callback to stop continuous recognition when an event is received
:param evt: event
"""
print('CLOSING on {}'.format(evt))
speech_recognizer.stop_continuous_recognition()
nonlocal done
done = True
def formatted_json(evt):
res = evt.result.json
return f'this is the result: {res}'
# Signal for events containing intermediate recognition results
if show == "all":
speech_recognizer.recognizing.connect(lambda evt: print('RECOGNIZING: {}'.format(evt)))
# Signal for events containing final recognition results (indicating a successful recognition attempt)
# original snippet from azure examples
#speech_recognizer.recognized.connect(lambda evt: print('RECOGNIZED: {}'.format(evt)))
# test method, to see all the json-fields
speech_recognizer.recognized.connect(lambda evt: print('RECOGNIZED: {}'.format(formatted_json(evt))))
# Signal for events indicating the start of a recognition session (operation).
speech_recognizer.session_started.connect(lambda evt: print('SESSION STARTED: {}'.format(evt)))
# Signal for events indicating the end of a recognition session (operation).
speech_recognizer.session_stopped.connect(lambda evt: print('SESSION STOPPED {}'.format(evt)))
# Signal for events containing canceled recognition results (indicating a recognition attempt that was canceled
# as a result or a direct cancellation request or, alternatively, a transport or protocol failure)
speech_recognizer.canceled.connect(lambda evt: print('CANCELED {}'.format(evt)))
speech_recognizer.session_stopped.connect(stop_cb)
speech_recognizer.canceled.connect(stop_cb)
# Start the actual transcription job
speech_recognizer.start_continuous_recognition()
while not done:
time.sleep(.5)
def from_file(file_path, key):
speech_config = speechsdk.SpeechConfig(subscription=key, region="eastus")
audio_input = speechsdk.AudioConfig(filename=file_path)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
return result.text.split(" ")
async def transcribe_file(speechFileFullPath, lang, fileName, writer):
"""Transcribe the given audio file."""
result = None
speech_config = speechsdk.SpeechConfig(subscription=Subscription,
region=Region)
speech_config.endpoint_id = Endpoint_id
speech_config.output_format = speechsdk.OutputFormat.Detailed
audio_input = speechsdk.AudioConfig(filename=speechFileFullPath)
speech_config.set_profanity(profanity_option=speechsdk.ProfanityOption.Raw)
#speech_config.speech_recognition_language=lang
recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
#auto_detect_source_language_config=auto_detect_source_language_config)
try:
result = recognizer.recognize_once_async().get()
confidence = 0
#temp = json.load()
for key in result.properties.keys():
temp = eval(result.properties[key])
if (type(temp) == dict):
confidence = temp["NBest"][0]["Confidence"]
#print(temp["NBest"][0]["Confidence"])
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
#print(fileName)
temp = {
"Filename": fileName,
"Culture/Accent": lang,
"Text": result.text,
"Confidence": confidence
}
recognition_config.recognitionResults.append(temp)
temp = None
#print(recognition_config.recognitionResults)
elif result.reason == speechsdk.ResultReason.NoMatch:
print("No speech could be recognized: {}".format(
result.no_match_details))
pass
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
print("Speech Recognition canceled: {}".format(
cancellation_details.reason))
if cancellation_details.reason == speechsdk.CancellationReason.Error:
print("Error details: {}".format(
cancellation_details.error_details))
except Exception as e:
print("Error in Transcribe Function : " + str(e))
return None
"""
def from_file():
speech_config = speechsdk.SpeechConfig(
subscription="<paste-your-subscription-key>",
region="<paste-your-region>")
audio_input = speechsdk.AudioConfig(filename="your_file_name.wav")
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
print(result.text)
def transcribe_azure(file_path):
"""performs one-shot speech recognition with input from an audio file"""
audio_input = speechsdk.AudioConfig(filename=file_path)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=AZURE_SPEECH_CONFIG,
audio_config=audio_input,
language='pt-BR')
result = speech_recognizer.recognize_once_async().get()
return result.text
def from_file(file):
speech_config = speechsdk.SpeechConfig(
subscription="f9769ca0076f491e91cbd90a1c6ed97c", region="eastus")
# audio_input = speechsdk.AudioConfig(filename=r"/Users/garyge/Desktop/NLP/final_project/audio_f0001/f0001_us_f0001_00001.wav")
audio_input = speechsdk.AudioConfig(filename=file)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
return result.text
def A2T(wav_file):
speech_config = speechsdk.SpeechConfig(subscription="<API_KEY>",
region="eastus")
audio_input = speechsdk.AudioConfig(filename=wav_file)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
speech_config.enable_dictation()
done = False
def stop_cb(evt):
print('CLOSING on {}'.format(evt))
speech_recognizer.stop_continuous_recognition()
nonlocal done
done = True
all_results = []
def handle_final_result(evt):
all_results.append(evt.result.text)
speech_recognizer.recognized.connect(handle_final_result)
speech_recognizer.recognizing.connect(
lambda evt: print('RECOGNIZING: {}'.format(evt)))
speech_recognizer.recognized.connect(
lambda evt: print('RECOGNIZED: {}'.format(evt)))
speech_recognizer.session_started.connect(
lambda evt: print('SESSION STARTED: {}'.format(evt)))
speech_recognizer.session_stopped.connect(
lambda evt: print('SESSION STOPPED {}'.format(evt)))
speech_recognizer.canceled.connect(
lambda evt: print('CANCELED {}'.format(evt)))
speech_recognizer.session_stopped.connect(stop_cb)
speech_recognizer.canceled.connect(stop_cb)
speech_recognizer.start_continuous_recognition()
while not done:
time.sleep(.5)
#print(all_results)
# Function to convert
def listToString(s):
# initialize an empty string
str1 = " "
# return string
return (str1.join(s))
return listToString(all_results)
"""with open(wav_file + ".txt",'w') as f:
def audiototext(self, audio_filename):
# Creates an audio configuration that points to an audio file.
audio_input = speechsdk.AudioConfig(filename=audio_filename)
# Creates a recognizer with the given settings
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=self.speech_config, audio_config=audio_input)
print("[Info] Prcoessing Audio to Text")
translated_text = self.recognizespeech(speech_recognizer)
return translated_text
def transcribe_audio_file_path(self, audio_file_path):
# For now supports wav, not mp3
# https://stackoverflow.com/questions/51614216/what-audio-formats-are-supported-by-azure-cognitive-services-speech-service-ss?rq=1
audio_config = speechsdk.AudioConfig(use_default_microphone=False,
filename=audio_file_path)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=self.speech_config, audio_config=audio_config)
done = False
transcript = ""
cancellation_details = None
def stop_cb(evt):
"""callback that stops continuous recognition upon receiving an event `evt`"""
print("CLOSING on {}".format(evt))
speech_recognizer.stop_continuous_recognition()
nonlocal done
done = True
def return_transcript(evt):
"""recognition is continuous, that is every sentence gets recognized separately.
We want to concatenate all the sentences and return the full transcript"""
nonlocal transcript
transcript += " "
transcript += evt.result.text
def return_cancellation_details(evt):
"""return cancellation details"""
nonlocal cancellation_details
cancellation_details = evt.result.cancellation_details.error_details
# Connect callbacks to the events fired by the speech recognizer
speech_recognizer.recognized.connect(return_transcript)
speech_recognizer.canceled.connect(return_cancellation_details)
# stop continuous recognition on either session stopped or canceled events
speech_recognizer.session_stopped.connect(stop_cb)
speech_recognizer.canceled.connect(stop_cb)
# Start continuous speech recognition
speech_recognizer.start_continuous_recognition()
while not done:
time.sleep(0.5)
if cancellation_details:
raise exceptions.Canceled("Azure Speech cancellation error: " +
cancellation_details)
if transcript == "":
raise exceptions.BlankTranscript(
"Azure Speech returned blank transcript")
return transcript
def speech_from_file(file):
if file[-4:] != ".wav":
print("error")
return
# return JsonResponse({"error": "file type error"})
speech_config = speechsdk.SpeechConfig(subscription=api_keys["microsoft-speech"]["key"], region=api_keys["microsoft-speech"]["region"])
audio_input = speechsdk.AudioConfig(filename=file)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
# return JsonResponse(result.text)
print(result.text)
return result.text
def snip_transcribe(output_list,
filename,
output_folder=output_folder,
speech_key=speech_key,
service_region=service_region):
speech_config = speechsdk.SpeechConfig(subscription=speech_key,
region=service_region)
speech_config.enable_dictation
def recognized_cb(evt):
if evt.result.reason == speechsdk.ResultReason.RecognizedSpeech:
# Do something with the recognized text
output_list[ind]['text'] = output_list[ind]['text'] + \
str(evt.result.text)
print(evt.result.text)
for ind, diag in enumerate(output_list):
t1 = diag['start_time']
t2 = diag['end_time']
newAudio = AudioSegment.from_wav(filename)
chunk = newAudio[t1 * 1000:t2 * 1000]
filename_out = output_folder + f"snippet_{diag['sequence_id']}.wav"
# Exports to a wav file in the current path.
chunk.export(filename_out, format="wav")
done = False
def stop_cb(evt):
"""callback that signals to stop continuous recognition upon receiving an event `evt`"""
print('CLOSING on {}'.format(evt))
nonlocal done
done = True
audio_input = speechsdk.AudioConfig(filename=filename_out)
speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=speech_config, audio_config=audio_input)
output_list[ind]['snippet_path'] = filename_out
speech_recognizer.recognized.connect(recognized_cb)
speech_recognizer.session_stopped.connect(stop_cb)
speech_recognizer.canceled.connect(stop_cb)
# Start continuous speech recognition
speech_recognizer.start_continuous_recognition()
while not done:
time.sleep(.5)
speech_recognizer.stop_continuous_recognition()
return output_list
def prepare_speech_recognizer(args):
hint = "See https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/get-started-speech-to-text"
subscription = read_key("speech-subscription", hint=hint)
print("ASR Language: %s" % (args.lang, ))
speech_config = speechsdk.SpeechConfig(
subscription=subscription,
region=args.region,
speech_recognition_language=args.lang,
)
input_stream = get_input_stream(args.port)
audio_config = speechsdk.AudioConfig(stream=input_stream, )
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_config)
return speech_recognizer
def from_file(base='', filename='default'):
speech_config = speechsdk.SpeechConfig(subscription="c4ac7a15c7204203b5179d7e745c238a", region="centralus")
wav_path = base + filename # wav 路径
blob_file_name = filename.split('.')[0] + '.txt' # txt文件名
txt_path = base + blob_file_name # txt文件路径
audio_input = speechsdk.AudioConfig(filename=wav_path)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
with open(txt_path, 'w', encoding="UTF-8") as fp:
fp.write(result.text)
# 进行备份
block_blob_service.create_blob_from_path(container_name, filename, wav_path)
block_blob_service.create_blob_from_path(container_name, blob_file_name, txt_path)
def recognize_audio(output, speech_key, service_region, language, filename, recognize_time=100):
"""
wav形式のデータから文字を起こす関数
---------------------------
Parameters
output: str
音声から起こしたテキスト（再帰的に取得する）
speech_key: str
Azure Speech SDKのキー
service_region: str
Azure Speech SDKのリージョン名
language: str
音声解析するための言語を指定する
filename: str
音声ファイルのパス
recognize_time: int
音声認識にかける時間
180秒の音声ファイルであれば100秒程度で十分
"""
# Azure Speech Config
speech_config = speechsdk.SpeechConfig(subscription=speech_key, region=service_region)
speech_config.enable_dictation()
# Language Setting
if language == '日本語':
speech_config.speech_recognition_language = "ja-JP"
elif language == '英語':
speech_config.speech_recognition_language = "en-US"
# Recognizing
audio_input = speechsdk.AudioConfig(filename=filename)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input)
def recognized(evt):
nonlocal output
output += evt.result.text
speech_recognizer.recognized.connect(recognized)
speech_recognizer.start_continuous_recognition()
time.sleep(recognize_time)
return output
def get_text_from_input(input_audio_filename, speech_config):
# Creates an audio configuration that points to an audio file.
# Replace with your own audio filename.
audio_input = speechsdk.AudioConfig(filename=input_audio_filename)
# Creates a recognizer with the given settings
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
print("Recognizing first result...")
# Starts speech recognition, and returns after a single utterance is recognized. The end of a
# single utterance is determined by listening for silence at the end or until a maximum of 15
# seconds of audio is processed. The task returns the recognition text as result.
# Note: Since recognize_once() returns only a single utterance, it is suitable only for single
# shot recognition like command or query.
# For long-running multi-utterance recognition, use start_continuous_recognition() instead.
result = speech_recognizer.recognize_once()
return result.text
def speech_to_text(audio_file):
print(audio_file)
audio_input = speechsdk.AudioConfig(filename=audio_file)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
print("Recognized: {}".format(result.text))
return result.text
elif result.reason == speechsdk.ResultReason.NoMatch:
print(f"No speech could be recognized: {result.no_match_details}")
return "ERROR_Rec"
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
print(f"Speech Recognition canceled: {cancellation_details.reason}")
if cancellation_details.reason == speechsdk.CancellationReason.Error:
print(f"Error details: {cancellation_details.error_details}")
return "ERROR_Canc"
return
async def handle_post(request):
reader = await request.multipart()
field = await reader.next()
assert field.name == 'audio'
filename = field.filename
size = 0
with open(os.path.join('.', filename), 'wb') as f:
while True:
chunk = await field.read_chunk()
if not chunk:
break
size += len(chunk)
f.write(chunk)
wavefile = filename + '.wav'
if os.path.exists(wavefile):
os.remove(wavefile)
ff = FFmpeg(inputs={filename: None}, outputs={wavefile: '-ac 1'})
ff.run()
speech_key, service_region = 'yourkey', 'regionsuchaswestus'
speech_config = speechsdk.SpeechConfig(subscription=speech_key,
region=service_region)
audio_config = speechsdk.AudioConfig(filename=wavefile)
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_config)
result = speech_recognizer.recognize_once()
res = 'No response'
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
res = 'Recognized: {}'.format(result.text)
elif result.reason == speechsdk.ResultReason.NoMatch:
res = 'No speech could be recognized: {}'.format(
result.no_match_details)
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
res = 'Speech Recognition canceled: {}'.format(
cancellation_details.reason)
# if cancellation_details.reason == speechsdk.CancellationReason.Error:
# print("Error details: {}".format(cancellation_details.error_details))
return web.Response(text=res)
def from_file():
f = open('output.txt', 'w')
d = open('result.txt', 'w')
speech_config = speechsdk.SpeechConfig(
subscription="18b27d3bc19143fe9d581e9b40b4253e", region="eastus")
audio_input = speechsdk.AudioConfig(filename="test.wav")
speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config,
audio_config=audio_input)
result = speech_recognizer.recognize_once_async().get()
d.write(result.text)
d.close()
returnedStuff = grammar_check.model_call(result.text)
for i in returnedStuff:
print(i + "\n")
f.write(i)
f.close()
def __init__(self, speech_key: str, service_region: str,
recognition_language: str, mic_id: str,
dict_mode_active: bool, callbackClass):
self.speech_recognizer = None
self.callbackClass = callbackClass
self.conversation = {}
# Setup Azure Speech Recognizer
if mic_id is not None:
audio_config = speechsdk.AudioConfig(device_name=mic_id)
else:
audio_config = None
speech_config = speechsdk.SpeechConfig(subscription=speech_key,
region=service_region)
speech_config.speech_recognition_language = recognition_language
if (dict_mode_active == True):
speech_config.enable_dictation()
print("Dictiation Mode enabled.")
self.speech_recognizer = speechsdk.SpeechRecognizer(
speech_config=speech_config, audio_config=audio_config)
# Connect callbacks to the events fired by the speech recognizer
self.speech_recognizer.session_started.connect(
lambda evt: print('SESSION STARTED: {}'.format(evt)))
self.speech_recognizer.session_stopped.connect(
lambda evt: print('SESSION STOPPED {}'.format(evt)))
self.speech_recognizer.recognizing.connect(
self.cb_recogonizing
) # every utterance, even when recogn. is not finished
self.speech_recognizer.recognized.connect(self.cb_recognized)
self.speech_recognizer.canceled.connect(self.cb_cancelled)
# stop continuous recognition on either session stopped or canceled events
self.speech_recognizer.session_stopped.connect(self.stop_cb)
self.speech_recognizer.canceled.connect(self.stop_cb)
# Improve recognition accuracy with custom lists
self.update_grammar_list(self.speech_recognizer)