Transcribe audio

audio-searcher.py

speech_recognizer = sr.Recognizer()
tokenizer = RegexpTokenizer(r'\w+')

def transcribe_audio(path):

    audio = AudioSegment.from_wav(path)
    file_name = os.path.basename(path) 

    chunks = split_on_silence(audio,
        min_silence_len = 600,
        silence_thresh = audio.dBFS-14,
        keep_silence=600,
    )
    audio_folder = "audio-chunks"
    if not os.path.isdir(audio_folder):
        os.mkdir(audio_folder)

    audio_text = ""

    for i, audio_chunk in enumerate(chunks, start=1):
        chunk_filename = os.path.join(audio_folder, f"{file_name}{i}")
        audio_chunk.export(chunk_filename, format="wav")

        with sr.AudioFile(chunk_filename) as source:
            audio_file = speech_recognizer.record(source)
            try:
                text = speech_recognizer.recognize_google(audio_file)

            except sr.UnknownValueError as e:
                print("An error has occurred while recognising audio file:", e)
            else:
                text = f"{text.capitalize()}. "
                print(chunk_filename, ":", text)
                audio_text += text

                os.remove(chunk_filename)

    return audio_text