app.py

import os
import spaces
import gradio as gr
from scipy.io.wavfile import write

@spaces.GPU()
def inference(audio):
  os.makedirs("out", exist_ok=True)
  write('test.wav', audio[0], audio[1])
  os.system("python3 -m demucs.separate -n htdemucs --two-stems=vocals -d cpu test.wav -o out")
  return "./out/htdemucs/test/vocals.wav","./out/htdemucs/test/no_vocals.wav"
    
title = "# Ilaria UVR 💖"
description = "Drag and drop an audio file to easily separate it! [Join AI Hub Discord Server](https://discord.gg/aihub).</p>"
article = "Made with 💖 by Ilaria"

examples = [['test.mp3']]

with gr.Blocks(title="Ilaria UVR 💖") as demo:
    gr.Markdown(f"# {title}")
    gr.Markdown(description)
    with gr.Row():
        audio_input = gr.Audio(type="numpy", label="Song")
    with gr.Row():
        vocals_output = gr.Audio(type="filepath", label="Vocals")
    with gr.Row():
        instrumentals_output = gr.Audio(type="filepath", label="Instrumentals")
    with gr.Row():
        gr.Examples(examples, inputs=audio_input)
    with gr.Row():
        btn = gr.Button("Separate Vocals")
    with gr.Row():
    gr.Markdown(article)
        btn.click(
            inference,
            inputs=[
                audio_input,
            ],
            outputs=[
                vocals_output, 
                instrumentals_output,
            ],
        )

demo.launch()