updated

app.py

@@ -3,7 +3,6 @@ import torch
 import gc
 import numpy as np
 import random
-import tempfile
 import os
 os.environ['ELASTIC_LOG_LEVEL'] = 'DEBUG'
 from transformers import AutoProcessor, pipeline
@@ -25,7 +24,6 @@ def cleanup_gpu():
         torch.cuda.synchronize()
         gc.collect()
 
-# Global variables for model caching with thread lock
 _generator = None
 _processor = None
 
@@ -115,20 +113,16 @@ def generate_music(text_prompt, duration=10, guidance_scale=3.0):
         print(f"[GENERATION] Audio shape: {audio_data.shape}")
         print(f"[GENERATION] Sample rate: {sample_rate}")
         
-        # Fix audio format for Gradio display
         if len(audio_data.shape) > 1:
             # If stereo or multi-channel, take first channel
             audio_data = audio_data[0] if audio_data.shape[0] < audio_data.shape[1] else audio_data[:, 0]
         
-        # Ensure it's 1D
         audio_data = audio_data.flatten()
         
-        # Normalize audio to prevent clipping
         max_val = np.max(np.abs(audio_data))
         if max_val > 0:
             audio_data = audio_data / max_val * 0.95  # Scale to 95% to avoid clipping
         
-        # Convert to float32 for Gradio
         audio_data = audio_data.astype(np.float32)
         
         print(f"[GENERATION] Final audio shape: {audio_data.shape}")
@@ -139,7 +133,7 @@ def generate_music(text_prompt, duration=10, guidance_scale=3.0):
     except Exception as e:
         print(f"[ERROR] Generation failed: {str(e)}")
         cleanup_gpu()
-        return None, None
+        return None
 
 
 with gr.Blocks(title="MusicGen Large - Music Generation") as demo:
@@ -177,7 +171,7 @@ with gr.Blocks(title="MusicGen Large - Music Generation") as demo:
         with gr.Column():
             audio_output = gr.Audio(
                 label="Generated Music",
-                type="numpy",
+                type="filepath",
                 format="wav",
                 interactive=False
             )
@@ -197,7 +191,6 @@ with gr.Blocks(title="MusicGen Large - Music Generation") as demo:
         show_progress=True
     )
 
-    # Example prompts - only text prompts now
     gr.Examples(
         examples=[
             "A groovy funk bassline with a tight drum beat",