app.py

import gradio as gr
import scipy
from diffusers import DiffusionPipeline

model_id = "harmonai/maestro-150k"
pipeline = DiffusionPipeline.from_pretrained(model_id)
pipeline = pipeline.to("cuda")

def denoise(length_sec):
    audios = pipeline(audio_length_in_s=length_sec).audios
    for audio in audios:
        scipy.io.wavfile.write("maestro_test.wav", pipe.unet.sample_rate, audio.transpose())
    return "maestro_test.wav"

gr.Interface(fn=denoise, inputs=gr.Slider(1.0, 6.0, value=3.0, step=0.5, label="Audio length in seconds"), outputs="audio").launch()