Update app.py

app.py

@@ -86,45 +86,57 @@ def generate_audio(prompt: str, audio_length: int, mg_model, mg_processor):
 # ---------------------------------------------------------------------
 # Gradio Interface
 # ---------------------------------------------------------------------
-def radio_imaging_app(user_prompt, llama_model_id, hf_token, audio_length):
+def radio_imaging_script(user_prompt, llama_model_id):
     # Load Llama 3 Pipeline with Zero GPU
     pipeline_llama = load_llama_pipeline_zero_gpu(llama_model_id, hf_token)
     if isinstance(pipeline_llama, str):
-        return pipeline_llama, None
+        return pipeline_llama
 
     # Generate Script
     script = generate_script(user_prompt, pipeline_llama)
+    return script
 
+def radio_imaging_audio(script, audio_length):
     # Load MusicGen
     mg_model, mg_processor = load_musicgen_model()
     if isinstance(mg_processor, str):
-        return script, mg_processor
+        return mg_processor
 
     # Generate Audio
     audio_data = generate_audio(script, audio_length, mg_model, mg_processor)
-    if isinstance(audio_data, str):
-        return script, audio_data
-
-    return script, audio_data
+    return audio_data
 
 # ---------------------------------------------------------------------
 # Interface
 # ---------------------------------------------------------------------
 with gr.Blocks() as demo:
     gr.Markdown("# 🎧 AI Radio Imaging with Llama 3 + MusicGen (Zero GPU)")
-    user_prompt = gr.Textbox(label="Enter your promo idea", placeholder="E.g., A 15-second hype jingle for a morning talk show.")
-    llama_model_id = gr.Textbox(label="Llama 3 Model ID", value="meta-llama/Meta-Llama-3-70B")
-    hf_token = gr.Textbox(label="Hugging Face Token", type="password")
-    audio_length = gr.Slider(label="Audio Length (tokens)", minimum=128, maximum=1024, step=64, value=512)
-
-    generate_button = gr.Button("Generate Promo Script and Audio")
-    script_output = gr.Textbox(label="Generated Script")
-    audio_output = gr.Audio(label="Generated Audio", type="filepath")
-
-    generate_button.click(
-        fn=radio_imaging_app,
-        inputs=[user_prompt, llama_model_id, hf_token, audio_length],
-        outputs=[script_output, audio_output]
-    )
+
+    # Script Generation Section
+    with gr.Box():
+        gr.Markdown("## Step 1: Generate the Promo Script")
+        user_prompt = gr.Textbox(label="Enter your promo idea", placeholder="E.g., A 15-second hype jingle for a morning talk show.")
+        llama_model_id = gr.Textbox(label="Llama 3 Model ID", value="meta-llama/Meta-Llama-3-70B")
+        generate_script_button = gr.Button("Generate Promo Script")
+        script_output = gr.Textbox(label="Generated Script", interactive=False)
+
+        generate_script_button.click(
+            fn=radio_imaging_script,
+            inputs=[user_prompt, llama_model_id],
+            outputs=script_output
+        )
+
+    # Audio Generation Section
+    with gr.Box():
+        gr.Markdown("## Step 2: Generate the Sound")
+        audio_length = gr.Slider(label="Audio Length (tokens)", minimum=128, maximum=1024, step=64, value=512)
+        generate_audio_button = gr.Button("Generate Sound from Script")
+        audio_output = gr.Audio(label="Generated Audio", type="filepath")
+
+        generate_audio_button.click(
+            fn=radio_imaging_audio,
+            inputs=[script_output, audio_length],
+            outputs=audio_output
+        )
 
 demo.launch(debug=True)