Update app.py

app.py

@@ -935,7 +935,7 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
     
     # Gradio 6.0+ moved theme/css to launch()
     with gr.Blocks(title="VibeVoice ASR Demo") as demo:
-        gr.Markdown("# 🎙️ VibeVoice ASR Demo")
+        gr.Markdown("# 🎙️ VibeVoice ASR Demo(VibeVoice自動語音辨識)")
         gr.Markdown("Upload audio files or record from microphone to get speech-to-text transcription with speaker diarization.")
         gr.Markdown(f"**Model loaded from:** `{model_path}`")
         
@@ -956,7 +956,8 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
                 do_sample_checkbox = gr.Checkbox(
                     value=False,
                     label="Enable Sampling",
-                    info="Enable random sampling instead of deterministic decoding"
+                    info="Enable random sampling instead of deterministic decoding",
+                    visible=False
                 )
                 
                 with gr.Column(visible=False) as sampling_params:
@@ -966,7 +967,8 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
                         value=0.0,
                         step=0.1,
                         label="Temperature",
-                        info="0 = greedy, higher = more random"
+                        info="0 = greedy, higher = more random",
+                        visible=False
                     )
                     top_p_slider = gr.Slider(
                         minimum=0.0,
@@ -974,7 +976,8 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
                         value=1.0,
                         step=0.05,
                         label="Top-p (Nucleus Sampling)",
-                        info="1.0 = no filtering"
+                        info="1.0 = no filtering",
+                        visible=False
                     )
                 
                 # Repetition penalty (works with both greedy and sampling)
@@ -984,7 +987,8 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
                     value=1.0,
                     step=0.01,
                     label="Repetition Penalty",
-                    info="1.0 = no penalty, higher = less repetition (works with greedy & sampling)"
+                    info="1.0 = no penalty, higher = less repetition (works with greedy & sampling)",
+                    visible=False
                 )
                 
                 # Context information section
@@ -1098,24 +1102,24 @@ def create_gradio_interface(model_path: str, default_max_tokens: int = 8192, att
             queue=False
         )
         
-        # Add examples
-        gr.Markdown("## 📋 Instructions")
-        gr.Markdown(f"""
-        1. **Upload Audio**: Use the audio component to upload a file or record from microphone
-           - **Supported formats**: {', '.join(sorted(set([ext.lower() for ext in COMMON_AUDIO_EXTS])))}
-           - Optionally set **Start/End time** (seconds or hh:mm:ss) to clip before transcription
-        2. **Context Info (Optional)**: Provide context to improve transcription accuracy
-           - Add hotwords, proper nouns, speaker names, or technical terms
-           - One item per line or comma-separated
-           - Examples: "John Smith", "OpenAI", "machine learning"
-        3. **Adjust Parameters**: Configure generation parameters as needed
-        4. **Transcribe**: Click "Transcribe" to get results
-        5. **Review Results**: 
-           - **Raw Output**: View the model's original output
-           - **Audio Segments**: Play individual segments directly to verify accuracy
-        
-        **Audio Segments**: Each segment shows the time range, speaker ID, transcribed content, and an embedded audio player for immediate verification.
-        """)
+        # # Add examples
+        # gr.Markdown("## 📋 Instructions")
+        # gr.Markdown(f"""
+        # 1. **Upload Audio**: Use the audio component to upload a file or record from microphone
+        #    - **Supported formats**: {', '.join(sorted(set([ext.lower() for ext in COMMON_AUDIO_EXTS])))}
+        #    - Optionally set **Start/End time** (seconds or hh:mm:ss) to clip before transcription
+        # 2. **Context Info (Optional)**: Provide context to improve transcription accuracy
+        #    - Add hotwords, proper nouns, speaker names, or technical terms
+        #    - One item per line or comma-separated
+        #    - Examples: "John Smith", "OpenAI", "machine learning"
+        # 3. **Adjust Parameters**: Configure generation parameters as needed
+        # 4. **Transcribe**: Click "Transcribe" to get results
+        # 5. **Review Results**: 
+        #    - **Raw Output**: View the model's original output
+        #    - **Audio Segments**: Play individual segments directly to verify accuracy
+        
+        # **Audio Segments**: Each segment shows the time range, speaker ID, transcribed content, and an embedded audio player for immediate verification.
+        # """)
     
     return demo, custom_css