Update app.py

app.py

@@ -14,14 +14,11 @@ controlnet = ControlNetModel.from_pretrained("lauraibnz/midi-audioldm", torch_dt
 pipe = AudioLDMControlNetPipeline.from_pretrained("cvssp/audioldm-m-full", controlnet=controlnet, torch_dtype=torch_dtype)
 pipe = pipe.to(device)
 
-def predict(prompt, midi_file=None, audio_length_in_s=5, controlnet_conditioning_scale=1.0, num_inference_steps=20):
- if midi_file:
- midi_file = midi_file.name
- else:
- midi_file = "test.mid"
+def predict(midi_file, prompt, audio_length_in_s, controlnet_conditioning_scale, num_inference_steps=20):
+ midi_file = midi_file.name
  midi = PrettyMIDI(midi_file)
  audio = pipe(prompt, midi=midi, audio_length_in_s=audio_length_in_s, num_inference_steps=num_inference_steps, controlnet_conditioning_scale=float(controlnet_conditioning_scale))
  return (16000, audio.audios.T)
 
-demo = gr.Interface(fn=predict, inputs=["text", gr.UploadButton("Upload a MIDI file", file_types=[".mid"]), gr.Slider(0, 30, value=5, step=5, label="duration (seconds)"), gr.Slider(0.0, 1.0, value=1.0, step=0.1, label="conditioning scale")], outputs="audio")
+demo = gr.Interface(fn=predict, inputs=[gr.File("test.mid", file_types=[".mid"]), "text", gr.Slider(0, 30, value=5, step=5, label="duration (seconds)"), gr.Slider(0.0, 1.0, value=1.0, step=0.1, label="conditioning scale")], outputs="audio")
 demo.launch()