Speech recognition

from io import BytesIO
from pathlib import Path
from pydub import AudioSegment
from pydub.silence import split_on_silence
import speech_recognition as sr
from concurrent.futures import ThreadPoolExecutor, ProcessPoolExecutor


def save_in_buffer(chunk):
    buffer = BytesIO()
    chunk.export(buffer, 'wav')
    buffer.seek(0)
    return buffer


def process_chapter(counter):
    chapter_nr = str(counter).zfill(2)
    chapter = AudioSegment.from_mp3(Path(rf'D:\Windows\Pulpit\Big data\{chapter_nr}_Big data.mp3'))
    chunks = split_on_silence(chapter, keep_silence=500, silence_thresh=-40)

    r = sr.Recognizer()

    def load_audio(buffer):
        with sr.AudioFile(buffer) as source:
            audio = r.record(source)
        return audio

    def recognize_text(audio):
        try:
            return r.recognize_google(audio, language='pl-PL')
        except sr.UnknownValueError:
            return '!-NotRecognized-!'
        except sr.RequestError as e:
            return '!-RequestError-!'

    audio_chunks = map(load_audio, map(save_in_buffer, chunks))

    with ThreadPoolExecutor(10) as pool:
        recognized_texts = pool.map(recognize_text, audio_chunks)

    recognized_texts = ' '.join(recognized_texts)

    with open(Path(f'results_{chapter_nr}.txt'), mode='w', encoding='utf-8') as file:
        file.write(recognized_texts)


if __name__ == '__main__':
    with ProcessPoolExecutor(4) as p:
        for i in range(2, 11):
            p.submit(process_chapter, i)