audio_to_text.py

from transformers import WhisperProcessor, WhisperForConditionalGeneration
import torchaudio
import torch
import os
from pydub import AudioSegment

# Install the specific version of FFmpeg
os.system("apt-get update")
os.system("apt-get install -y ffmpeg=7:5.1.6-0+deb12u1")
# Verify the installed version of FFmpeg
os.system("ffmpeg -version")

# load model and processor
processor = WhisperProcessor.from_pretrained("openai/whisper-small")
model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-small")
model.config.forced_decoder_ids = None


def audio_to_text(webm_file_path):
    wav_file = "recorded_audio.wav"
    absolute_path = os.path.abspath(webm_file_path)

    # Load and convert audio
    # Check if the file exists
    if os.path.exists(webm_file_path):
        wav_audio = AudioSegment.from_file(absolute_path, format="webm")
        wav_audio.export(wav_file, format="wav")
        # Load the audio and resample it
        waveform, sample_rate = torchaudio.load('recorded_audio.wav')
        resampler = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)
        waveform = resampler(waveform)
        waveform = waveform.squeeze().numpy()
        input_features = processor(waveform, sampling_rate=16000, return_tensors="pt").input_features
        # generate token ids
        predicted_ids = model.generate(input_features)
        # decode token ids to text
        transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
        return transcription
    else:
        return None