transcribe_mp3.py

import os
import speech_recognition as sr
from pydub import AudioSegment
from pydub.silence import split_on_silence

r = sr.Recognizer()
r.energy_threshold = 300


def clean_text(text:str):
    text = text.strip()
    text = text[0].upper() + text[1:]
    return text + ". "

def split_and_transcribe(path):
    print("Opening audio segment...")
    sound = AudioSegment.from_mp3(path)
    print("Splitting audio file...")
    chunks = split_on_silence(
        sound, min_silence_len=1000, silence_thresh=sound.dBFS - 14, keep_silence=500
    )
    folder_name = "audio-chunks"

    if not os.path.isdir(folder_name):
        print("Creating a new temp directory for audio chunks.")
        os.mkdir(folder_name)

    document = ""

    for i, audio_chunk in enumerate(chunks, start=1):
        print(f"Exporting chunk {i}")
        chunk_filename = os.path.join(folder_name, f"chunk{i}.wav")
        audio_chunk.export(chunk_filename, format="wav")
        print("Opening chunk...")
        with sr.AudioFile(chunk_filename) as source:
            print(f"Recognizing chunk with name {chunk_filename}")
            audio_listened = r.record(source)
            print("Converting to text...")
            try:
                text: str = r.recognize_google(audio_listened)
            except sr.UnknownValueError as e:
                print("Unknown Value Error: ", str(e))
                if document and not document.endswith("\n"):
                    document += "\n\n"
            else:
                text: str = clean_text(text)
                print(f"{chunk_filename} =>  '{text}'")
                document += text
    
    return document 


def extractname(filepath):
    base = os.path.basename(filepath)
    base_split = base.split(".")[:-1]
    return ".".join(base_split)


def transcribe(filename, to_filename):
    print(f"Transcribing {filename} => {to_filename}")
    return split_and_transcribe(filename)


def replace_common_transcription_errors(text: str):
    """Replace AND and BUT sentence starts/ends, and other errors."""
    print("Replacing transcription errors...")
    text = text.replace("and.", "and...")
    text = text.replace("and... And", "and...")
    text = text.replace(". And ", ", and ")
    text = text.replace(". But ", ", but ")
    # text = text.replace(" i ", " I ")
    return text


def main():
    mp3s = []
    print("Finding all MP3s...")
    for file in os.listdir():
        if file.endswith("MP3"):
            mp3s.append(file)

    if len(mp3s) > 0:
        for file in mp3s:
            print(f"Processing MP3 '{file}'")
            filename = extractname(file)
            transcribedfile = f"{filename}.txt"
            if os.path.exists(transcribedfile):
                print("A transcription already exists: "+str(transcribedfile))
            else:
                print("No text file found. Transcribing...")
                text = transcribe(file, transcribedfile)
                text = replace_common_transcription_errors(text)
                print("\nGot Text:\n\n")
                print(text)
                print()
                with open(transcribedfile, "w") as file:
                    print("Writing to file...")
                    file.write(f"Full Transcript for Audio File {filename}.MP3\n\n")
                    file.write(text)
                    file.close()


main()