Update app.py

app.py

@@ -1,55 +1,63 @@
 import gradio as gr
-from audiocraft.models import MusicGen
+from transformers import pipeline
 import torch
 import numpy as np
 
-# 1. Load the model (Small is safest for 16GB RAM)
-model = MusicGen.get_pretrained('facebook/musicgen-small')
+# Load Tom's brain (MusicGen-Small)
+# We use 'cpu' because Free Spaces don't have GPUs
+synthesiser = pipeline("text-to-audio", "facebook/musicgen-small", device="cpu")
 
-# 2. Define the CSS to hide footer, API, and settings
+# Your requested Clean UI CSS
 custom_css = """
 footer {display: none !important;}
 .show-api {display: none !important;}
 button.reference {display: none !important;}
-#component-0 {margin-bottom: 24px;}
 """
 
-def generate_music(description, duration):
-    if not description:
+def tom_make_music(prompt, duration):
+    if not prompt:
         return None
-        
-    model.set_generation_params(duration=duration)
-    wav = model.generate([description])
     
-    # Process for Gradio output
-    wav = wav.cpu().numpy().squeeze()
-    sampling_rate = model.sample_rate
-    return (sampling_rate, wav)
+    # 50 tokens = ~1 second of audio
+    max_tokens = int(duration * 50)
+    
+    # Generate the audio
+    output = synthesiser(
+        prompt, 
+        forward_params={
+            "do_sample": True, 
+            "max_new_tokens": max_tokens
+        }
+    )
+    
+    # Return as (sampling_rate, audio_data)
+    return (output["sampling_rate"], output["audio"].squeeze())
 
-# 3. Build the Interface
-with gr.Blocks(css=custom_css) as demo:
-    gr.Markdown("# 🎵 Tom-The-Ai-Music-AI")
-    gr.Markdown("Describe your vibe and Tom will compose it. *Note: CPU generation takes about 60-90 seconds.*")
+# Build the Interface
+with gr.Blocks(css=custom_css, title="Tom The Music AI") as demo:
+    gr.Markdown("# 🎹 Tom-The-Ai-Music-AI")
+    gr.Markdown("I'm Tom. Describe a song and I'll generate it for you.")
     
     with gr.Row():
         with gr.Column():
-            prompt = gr.Textbox(
-                label="What should Tom play?", 
-                placeholder="e.g., Synthwave with heavy bass and 80s drums",
+            input_text = gr.Textbox(
+                label="What should I play?", 
+                placeholder="e.g., Chill lo-fi beats for studying",
                 lines=3
             )
-            duration = gr.Slider(minimum=2, maximum=15, value=8, step=1, label="Seconds")
-            btn = gr.Button("Generate Music", variant="primary")
-        
+            duration_slider = gr.Slider(
+                minimum=2, maximum=15, value=8, step=1, label="Seconds"
+            )
+            btn = gr.Button("Generate with Tom", variant="primary")
+            
         with gr.Column():
             output_audio = gr.Audio(label="Tom's Masterpiece")
 
-    btn.click(fn=generate_music, inputs=[prompt, duration], outputs=output_audio)
+    btn.click(fn=tom_make_music, inputs=[input_text, duration_slider], outputs=output_audio)
 
-# 4. Launch with the clean settings
 if __name__ == "__main__":
     demo.launch(
         server_name="0.0.0.0", 
         server_port=7860,
-        show_api=False  # Extra layer to hide API link
+        show_api=False
     )