app.py

try:
    import spaces  # Para ambientes específicos como Hugging Face Spaces
    USING_SPACES = True
except ImportError:
    USING_SPACES = False

import gradio as gr
import torch
import torchaudio
from functools import partial
from resemble_enhance.enhancer.inference import denoise, enhance

def gpu_decorator(func):
    if USING_SPACES:
        return spaces.GPU(func)
    else:
        return func

@gpu_decorator
def _fn(path, solver, nfe, tau, denoising, unlimited):
    if path is None:
        return "Please upload an audio file.", None
    info = torchaudio.info(path)
    if not unlimited and (info.num_frames / info.sample_rate > 60):
        return "Only audio files shorter than 60 seconds are supported.", None
    solver = solver.lower()
    nfe = int(nfe)
    lambd = 0.9 if denoising else 0.1
    dwav, sr = torchaudio.load(path)
    dwav = dwav.mean(dim=0)
    wav1, new_sr = denoise(dwav, sr)  # Remove o argumento device
    wav2, new_sr = enhance(dwav, sr, nfe=nfe, solver=solver, lambd=lambd, tau=tau)  # Remove o argumento device
    wav1 = wav1.cpu().numpy()
    wav2 = wav2.cpu().numpy()
    return (new_sr, wav1), (new_sr, wav2)



with gr.Blocks() as app:
    inputs = [
        gr.Audio(type="filepath", label="Input Audio"),
        gr.Dropdown(
            choices=["Midpoint", "RK4", "Euler"],
            value="Midpoint",
            label="CFM ODE Solver (Midpoint is recommended)",
        ),
        gr.Slider(
            minimum=1,
            maximum=128,
            value=64,
            step=1,
            label="CFM Number of Function Evaluations (higher values in general yield better quality but may be slower)",
        ),
        gr.Slider(
            minimum=0,
            maximum=1,
            value=0.5,
            step=0.01,
            label="CFM Prior Temperature (higher values can improve quality but can reduce stability)",
        ),
        gr.Checkbox(
            value=False,
            label="Denoise Before Enhancement (tick if your audio contains heavy background noise)",
        ),
        gr.Checkbox(
            value=False,
            label="Allow Unlimited Audio Length (supports files longer than 60 seconds)",
        ),
    ]
    outputs = [
        gr.Audio(label="Output Denoised Audio"),
        gr.Audio(label="Output Enhanced Audio"),
    ]
    # Interface Gradio
    interface = gr.Interface(
        fn=_fn,
        title="Audio Enhancement",
        description="AI-driven audio enhancement powered by Resemble AI.",
        inputs=inputs,
        outputs=outputs,
    )
    app = interface.queue()


def main():
    global app
    print("Starting app...")
    app.launch(share=True)


if __name__ == "__main__":
    if not USING_SPACES:
        main()
    else:
        app.launch()