app.py

import torch
import numpy as np
from transformers import pipeline, SpeechT5Processor, SpeechT5ForTextToSpeech, SpeechT5HifiGan
from datasets import load_dataset
import gradio as gr

# Configuración del dispositivo (GPU si está disponible)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Pipeline de traducción automática de voz
pipe = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=0 if torch.cuda.is_available() else -1)

def translate(audio):
    outputs = pipe(audio, generate_kwargs={"task": "translate", "max_new_tokens": 256})
    return outputs["text"]

# Modelos para síntesis de voz
processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
model = SpeechT5ForTextToSpeech.from_pretrained("microsoft/speecht5_tts")
vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")

model.to(device)
vocoder.to(device)

# Embedding del hablante
embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
speaker_embeddings = torch.tensor(embeddings_dataset[6000]["xvector"]).unsqueeze(0)

def synthesise(text):
    inputs = processor(text=text, return_tensors="pt")
    speech = model.generate_speech(
        inputs["input_ids"].to(device),
        speaker_embeddings.to(device),
        vocoder=vocoder
    )
    return speech.cpu()

# Conversión final
target_dtype = np.int16
max_range = np.iinfo(target_dtype).max

def speech_to_speech_translation(audio):
    translated_text = translate(audio)
    synthesised_speech = synthesise(translated_text)
    synthesised_speech = (synthesised_speech.numpy() * max_range).astype(np.int16)
    return 16000, synthesised_speech

# Interfaz Gradio
demo = gr.Interface(
    fn=speech_to_speech_translation,
    inputs=gr.Audio(sources="microphone", type="filepath"),
    outputs=gr.Audio(label="Generated Speech", type="numpy"),
)

demo.launch(debug=True)