app.py

import streamlit as st
import torch
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
from gtts import gTTS
import numpy as np
import sounddevice as sd

class VoiceRecognition:
    def __init__(self):
        self.processor = Wav2Vec2Processor.from_pretrained("mistralai/Mixtral-8x7B-Instruct-v0.1")
        self.model = Wav2Vec2ForCTC.from_pretrained("mistralai/Mixtral-8x7B-Instruct-v0.1")  
        self.sample_rate = 16000

    def listen(self):
        st.write("Escuchando...")
        audio_data = sd.rec(int(self.sample_rate * 5), samplerate=self.sample_rate, channels=1, dtype='float32')
        sd.wait()
        st.write("Grabación terminada.")
        return audio_data.flatten()

    def vad(self, audio):
        threshold = 0.02
        return audio[np.abs(audio) > threshold]

    def transcribe(self, audio):
        input_values = self.processor(audio, return_tensors="pt", sampling_rate=self.sample_rate).input_values
        with torch.no_grad():
            logits = self.model(input_values).logits
        predicted_ids = torch.argmax(logits, dim=-1)
        return self.processor.decode(predicted_ids[0])

    def text_to_speech(self, text):
        tts = gTTS(text=text, lang='es')
        output_path = "response.mp3"
        tts.save(output_path)
        return output_path

def main():
    st.title("Asistente de Voz - Reconocimiento de Voz")
    recognizer = VoiceRecognition()
    
    if st.button("Iniciar Grabación"):
        audio = recognizer.listen()
        audio_vad = recognizer.vad(audio)
        
        if audio_vad.size > 0:
            transcription = recognizer.transcribe(audio_vad)
            st.write(f"Texto transcrito: {transcription}")
            audio_path = recognizer.text_to_speech(transcription)
            st.audio(audio_path)
        else:
            st.write("No se detectó actividad de voz.")

if __name__ == "__main__":
    main()