keep audio component visible

app.py

@@ -830,11 +830,12 @@ with gr.Blocks(title="LLM Inference with ZeroGPU") as demo:
         # 1. Update UI to "generating" state.
         #    Crucially, we do NOT update the `chat` component here, as the backend
         #    will provide the correctly formatted history in the first response chunk.
+        #    Keep audio visible but clear it - Gradio will show loading state
         yield {
             txt: gr.update(value="", interactive=False),
             submit_btn: gr.update(interactive=False),
             cancel_btn: gr.update(visible=True),
-            tts_audio_output: gr.update(visible=False, value=None),  # Hide audio during generation
+            tts_audio_output: gr.update(value=None),  # Clear audio but keep visible
         }
 
         cancelled = False
@@ -849,9 +850,9 @@ with gr.Blocks(title="LLM Inference with ZeroGPU") as demo:
                     dbg: debug,
                 }
 
-                # Show audio output when audio is generated (final yield with TTS)
+                # Update audio output when audio is generated (final yield with TTS)
                 if audio is not None:
-                    update_dict[tts_audio_output] = gr.update(visible=True, value=audio)
+                    update_dict[tts_audio_output] = gr.update(value=audio)
 
                 yield update_dict
         except GeneratorExit:
@@ -890,7 +891,7 @@ with gr.Blocks(title="LLM Inference with ZeroGPU") as demo:
             txt: gr.update(interactive=True),
             submit_btn: gr.update(interactive=True),
             cancel_btn: gr.update(visible=False),
-            tts_audio_output: gr.update(visible=False, value=None),
+            tts_audio_output: gr.update(value=None),  # Clear audio but keep visible
         }
 
     # Event for submitting text via Enter key or Submit button
@@ -931,6 +932,6 @@ with gr.Blocks(title="LLM Inference with ZeroGPU") as demo:
     )
 
     # Clear chat action
-    clr.click(fn=lambda: ([], "", "", gr.update(visible=False, value=None)), outputs=[chat, txt, dbg, tts_audio_output])
+    clr.click(fn=lambda: ([], "", "", gr.update(value=None)), outputs=[chat, txt, dbg, tts_audio_output])
 
     demo.launch(theme=CUSTOM_THEME, css=CUSTOM_CSS)