Create transcript from MP3 file by using GCP Speech to text API

mp3_to_text.py

import os
import re
import wave
import tempfile
from pydub import AudioSegment
from google.cloud import speech
from google.cloud.speech import enums
from google.cloud.speech import types

WORK_DIR = tempfile.mkdtemp()

def mp3_to_text(fpath):
    """
    Text from a wave file by using GCP Speach to text API

    Params:
        (str) fpath: path to the mp3 file
    Returns:
        (str[]) recognized texts
    """
    wav_path = mp3_to_wav(fpath)
    return wav_to_text(wav_path)


def wav_to_text(fpath):
    """
    Text from a wave file by using GCP Speach to text API

    Params:
        (str) fpath: path to the mp3 file
    Returns:
        (str[]) recognized texts
    """
    frame_rate = get_frame_rate(fpath)

    client = speech.SpeechClient()
    with open(fpath, 'rb') as fp:
        content = fp.read()
        audio = types.RecognitionAudio(content=content)

    config = types.RecognitionConfig(
        encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
        sample_rate_hertz=frame_rate,
        language_code='en-US'
    )

    response = client.recognize(config, audio)
    
    text = []
    for result in response.results:
        text.append(result.alternatives[0].transcript)

    return text


def get_frame_rate(fpath):
    """
    Get frame rate of a wave file

    Params:
        (str) fpath: path to a wav file
    Returns:
        (int) the frame rate
    """
    frame_rate = None
    with wave.open(fpath, 'rb') as fp:
        frame_rate = fp.getframerate()

    return frame_rate


def mp3_to_wav(fpath):
    """
    Convert mp3 to wav and returns the path of the created file

    Params:
        (str) fpath: path to the mp3 file
    Returns:
        (str) Created wav file path
    """
    basename = os.path.basename(fpath)
    new_fpath = WORK_DIR + re.sub(r"\.mp3$", ".wav", basename)

    sound = AudioSegment.from_mp3(fpath)
    sound = sound.set_channels(1)
    sound.export(new_fpath, format="wav", parameters=["-ac", "1"])
    return new_fpath


if __name__ == "__main__":
    print(mp3_to_text("audio/002/2.mp3"))

Pipfile

[[source]]
name = "pypi"
url = "https://pypi.org/simple"
verify_ssl = true

[dev-packages]

[packages]
google-cloud-speech = "*"
pydub = "*"

[requires]
python_version = "3.7"

• https://cloud.google.com/speech-to-text/docs/reference/libraries
• https://towardsdatascience.com/how-to-use-google-speech-to-text-api-to-transcribe-long-audio-files-1c886f4eb3e9