interstellar_ai/py/voice.py

import io
import numpy as np
from faster_whisper import WhisperModel
from pydub import AudioSegment


class VoiceRecognition:
    @staticmethod
    def recognition(audio):
        audio_buffer = io.BytesIO(audio.read())
        
        try:
            audio_segment = AudioSegment.from_file(audio_buffer, format="ogg")
            
            wav_io = io.BytesIO()
            audio_segment.export(wav_io, format="wav")
            wav_io.seek(0)
        except:
            print("audio to wav failed")
        
        model_size = "base"
        model = WhisperModel(model_size, device="cpu", compute_type="int8")
        
        segments, _ = model.transcribe(wav_io)
        transcription = ""
        for segment in segments:
            transcription += segment.text + " "
        result = transcription.strip()
        print(result)
        return result

# npm install @ffmpeg/ffmpeg @ffmpeg/util @ffmpeg/types @ffmpeg/core-mt