voice_cloner.py

# voice_cloner.py
from transformers import BarkModel, AutoProcessor
import torchaudio
import torch
import os

def clone_and_generate_text(text, reference_audio_path, language="English", emotion="Neutral"):
    processor = AutoProcessor.from_pretrained("suno/bark")
    model = BarkModel.from_pretrained("suno/bark")

    device = "cuda" if torch.cuda.is_available() else "cpu"
    model.to(device)

    # Load and process reference audio
    speech_array, sampling_rate = torchaudio.load(reference_audio_path)
    speech_array = torchaudio.functional.resample(speech_array, sampling_rate, 16000)
    speech_array = speech_array.mean(dim=0).unsqueeze(0)  # mono

    inputs = processor(
        text=text,
        voice_preset="v2/en_speaker_9",  # generic fallback voice
        return_tensors="pt"
    ).to(device)

    with torch.no_grad():
        speech = model.generate(**inputs)

    output_path = "output_voice.wav"
    torchaudio.save(output_path, speech.cpu(), 22050)
    return output_path