interstellar_ai/py/voice.py

import io
import numpy as np
import whisper
from pydub import AudioSegment


class VoiceRecognition:
    @staticmethod
    def recognition(audio):
        audio_data = audio.read()
        with io.BytesIO(audio_data) as audio_buffer:
            audio_segment = AudioSegment.from_ogg(audio_buffer)
            
            raw_data = np.array(audio_segment.get_array_of_samples())
            
            if audio_segment.channels > 1:
                raw_data = raw_data.reshape((-1, audio_segment.channels))[:, 0]
            
            audio_data = raw_data.astype(np.float32) / (2 ** (audio_segment.sample_width * 8 - 1))
        
        model = whisper.load_model("base")
        result = model.transcribe(audio_data)
        print(result["text"])
        return result["text"]

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