app.py

import os
import time
import torch
import gradio as gr
from pathlib import Path
import torchaudio
from chatterbox.tts import ChatterboxTTS

# Initialize output directory
OUTPUT_DIR = Path("outputs")
OUTPUT_DIR.mkdir(exist_ok=True)

# Load model
model = ChatterboxTTS.from_pretrained(device="cpu")

def tts_and_save(text, ref_wav, exaggeration, temperature, seed, cfg_weight, progress=gr.Progress()):
    progress(0, desc="Starting...")
    if seed != 0:
        torch.manual_seed(int(seed))
    
    progress(0.2, desc="Processing text...")
    time.sleep(0.5)  # Simulate processing
    
    progress(0.5, desc="Generating audio...")
    wav = model.generate(
        text,
        audio_prompt_path=ref_wav,
        exaggeration=exaggeration,
        temperature=temperature,
        cfg_weight=cfg_weight,
    )
    
    progress(0.8, desc="Saving file...")
    timestamp = time.strftime("%Y%m%d_%H%M%S")
    filename = OUTPUT_DIR / f"tts_{timestamp}.wav"
    torchaudio.save(str(filename), wav.cpu(), model.sr)
    
    progress(1.0, desc="Completed!")
    return (model.sr, wav.squeeze(0).numpy()), str(filename)

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            text = gr.Textbox(
                label="Input Text",
                lines=10,
                max_lines=100,
                placeholder="Enter your text here (no character limit)..."
            )
            ref_wav = gr.Audio(label="Reference Audio (optional)", sources=["upload", "microphone"], type="filepath")
            
            with gr.Row():
                exaggeration = gr.Slider(0.25, 2, value=0.5, step=0.05, label="Exaggeration")
                cfg_weight = gr.Slider(0.2, 1, value=0.5, step=0.05, label="CFG/Pace")
            
            with gr.Row():
                temperature = gr.Slider(0.05, 5, value=0.8, step=0.05, label="Temperature")
                seed = gr.Number(value=0, label="Random Seed (0=random)", precision=0)
            
            btn = gr.Button("Generate and Auto-Save", variant="primary")
        
        with gr.Column():
            output_audio = gr.Audio(label="Audio Preview")
            saved_path = gr.Textbox(label="Saved Path", interactive=False)
            progress_bar = gr.Textbox(label="Progress", visible=False)  # Hidden as we're using gr.Progress()

    btn.click(
        tts_and_save,
        inputs=[text, ref_wav, exaggeration, temperature, seed, cfg_weight],
        outputs=[output_audio, saved_path]
    )

if __name__ == "__main__":
    demo.launch()