voice_handler.py

import sounddevice as sd
import numpy as np
import librosa
import torch
from transformers import pipeline

class VoiceHandler:
    def __init__(self):
        self.sample_rate = 16000
        self.emotion_classifier = pipeline("audio-classification", 
                                        model="ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition")

    def record_audio(self, duration=5):
        """Record audio for specified duration"""
        recording = sd.rec(int(duration * self.sample_rate),
                         samplerate=self.sample_rate,
                         channels=1)
        sd.wait()
        return recording

    def process_audio(self, audio_data):
        """Process audio and detect emotion"""
        # Convert to mono if needed
        if len(audio_data.shape) > 1:
            audio_data = np.mean(audio_data, axis=1)
        
        # Normalize audio
        audio_data = librosa.util.normalize(audio_data)
        
        # Get emotion
        emotion = self.emotion_classifier(audio_data)
        
        return audio_data, emotion[0]['label']

    def enhance_audio(self, audio_data):
        """Enhance audio quality"""
        # Noise reduction
        y = librosa.effects.preemphasis(audio_data)
        
        # Normalize
        y = librosa.util.normalize(y)
        
        return y