SpeechRecognition

💡

Speech Recognition 패키지에서 기본이 되는 google api와 한국어와 오프라인 모드를 지원하는 vosk, whisper를 중점적으로 stt 기본 코드를 활용

 

 

1. 개발환경 구성

pip3 install SpeechRecognition

# for MAC
brew install portaudio
pip3 install pyaudio

# for Ubuntu
sudo apt-get install python-pyaudio python3-pyaudio
sudo apt-get install portaudio19-dev python-all-dev python3-all-dev
sudo pip install pyaudio

# for api
python3 -m pip install vosk
python3 -m pip install git+https://github.com/openai/whisper.git soundfile

 

 

2. 예제 코드

1. Google API

• 코드

  import speech_recognition as sr
  
  r = sr.Recognizer()
  with sr.Microphone() as source:
      print('listening...')
      audio = r.listen(source, timeout=10, phrase_time_limit=10)
      print("......")
  
  try:
      text = r.recognize_google(audio, language='ko')
      print(text)
  except sr.UnknownValueError:
      print("Recognizer Failed..")
  except sr.RequestError as e:
      print("Request Failed...", e)

 

 

2. Vosk

• 환경 구성
  • 모델 다운로드
    • https://alphacephei.com/vosk/models
    • https://alphacephei.com/vosk/models/vosk-model-small-ko-0.22.zip
  • 모델 파일 다운로드 이후에, 프로젝트 폴더 하위에 model 폴더 생성 후 압축해제한다.

    

• 코드

  import speech_recognition as sr
  
  r = sr.Recognizer()
  with sr.Microphone() as source:
      print('listening...')
      audio = r.listen(source, timeout=10, phrase_time_limit=10)
      print("......")
  
  try:
      text = r.recognize_vosk(audio, language='ko')
      print(text)
  except sr.UnknownValueError:
      print("Recognizer Failed..")
  except sr.RequestError as e:
      print("Request Failed...", e)

 

 

3. whisper

• 코드

  import speech_recognition as sr
  
  r = sr.Recognizer()
  with sr.Microphone() as source:
      print('listening...')
      audio = r.listen(source, timeout=10, phrase_time_limit=10)
      print("......")
  
  try:
      text = r.recognize_whisper(audio, language='ko')
      print(text)
  except sr.UnknownValueError:
      print("Recognizer Failed..")
  except sr.RequestError as e:
      print("Request Failed...", e)

 

 

 

3. 테스트

• 테스트 파일
  • 안녕하세요. 이것은 테스트 문장입니다.

• 코드

  import speech_recognition as sr
  import json
  
  r = sr.Recognizer()
  with sr.Microphone() as source:
      print('listening...')
      audio = r.listen(source, timeout=10, phrase_time_limit=10)
      print("......")
  
  try:
      text_google = r.recognize_google(audio, language='ko', show_all=True)
      text_google = dict(text_google)['alternative'][0]['transcript'] if 'alternative' in dict(text_google).keys() else ""
      text_vosk = r.recognize_vosk(audio, language='ko')
      text_vosk = json.loads(text_vosk)['text']
      text_whisper = r.recognize_whisper(audio, language='ko')
      print("[Google]", text_google)
      print("[Vosk]", text_vosk)
      print("[whisper]", text_whisper)
  except sr.UnknownValueError:
      print("Recognizer Failed..")
  except sr.RequestError as e:
      print("Request Failed...", e)

• 인식 결과

  [Google] 안녕하세요 이것은 테스트 문장입니다
  [Vosk] 륜 리아의 아이버슨 테스트 문자 입니다
  [whisper]  안녕하세요 이것은 테스트 문장입니다

  

 

 

 

참고

SpeechRecognition

Library for performing speech recognition, with support for several engines and APIs, online and offline.

https://pypi.org/project/SpeechRecognition/