Update app.py

app.py

@@ -29,8 +29,50 @@ code {
     color: #b45309 !important;
     font-weight: 600;
 }
+
+.audio-warning {
+    color: #ff6b35 !important;
+    font-weight: 600;
+    margin: 10px 0;
+}
+
+.audio-error {
+    color: #dc2626 !important;
+    font-weight: 600;
+    margin: 10px 0;
+}
 """
 
+def validate_audio_duration(audio_data):
+    """
+    Validate audio duration and return appropriate message
+    Returns: (is_valid, warning_message)
+    """
+    if audio_data is None:
+        return True, ""
+    
+    sample_rate, audio_array = audio_data
+    duration_seconds = len(audio_array) / sample_rate
+    
+    if duration_seconds > 10:
+        error_msg = f"""
+        <div class="audio-error">
+        ❌ Error: Audio is {duration_seconds:.1f} seconds long. Maximum allowed is 10 seconds.<br>
+        ❌ エラー: 音声が{duration_seconds:.1f}秒です。最大10秒まで許可されています。
+        </div>
+        """
+        return False, error_msg
+    elif duration_seconds > 8:
+        warning_msg = f"""
+        <div class="audio-warning">
+        ⚠️ Warning: Your audio is {duration_seconds:.1f} seconds, it will eat up precious context and may result in poor generation.<br>
+        ⚠️ 警告: 音声が{duration_seconds:.1f}秒を超えています。貴重なコンテキストを消費し、生成品質が低下する可能性があります。
+        </div>
+        """
+        return True, warning_msg
+    else:
+        return True, ""
+
 def load_examples(csv_path):
     examples = []
     
@@ -95,9 +137,17 @@ def run_generation_pipeline_client(
     top_k,
     temperature,
     use_chained_longform,
-    seed  # Add seed parameter
+    seed,  # Add seed parameter
+    audio_warning_display
 ):
     try:
+        # Validate audio duration first
+        is_valid, warning_msg = validate_audio_duration(audio_prompt)
+        
+        if not is_valid:
+            # Return error without processing
+            return None, "Status: Audio too long. Please use audio under 10 seconds."
+        
         # Handle audio prompt - save to temporary file if provided
         audio_prompt_for_api = None
         
@@ -173,9 +223,17 @@ def run_duration_generation_pipeline_client(
     add_steps,
     use_duration_aware,
     chars_per_second,
-    seed  # Add seed parameter
+    seed,  # Add seed parameter
+    audio_warning_display_dur
 ):
     try:
+        # Validate audio duration first
+        is_valid, warning_msg = validate_audio_duration(audio_prompt)
+        
+        if not is_valid:
+            # Return error without processing
+            return None, "Status: Audio too long. Please use audio under 10 seconds."
+        
         # Handle audio prompt - save to temporary file if provided
         audio_prompt_for_api = None
         
@@ -242,6 +300,15 @@ def run_duration_generation_pipeline_client(
     except Exception as e:
         return None, f"Status: Connection error: {str(e)}"
 
+# Audio validation callback
+def on_audio_upload(audio_data):
+    """Validate audio when uploaded and return warning message"""
+    is_valid, warning_msg = validate_audio_duration(audio_data)
+    if not is_valid:
+        # Clear the audio input if it's too long
+        return None, warning_msg
+    return audio_data, warning_msg
+
 # Load examples
 examples_csv_path = "./samples.csv"  # Adjust path as needed for client side
 example_list = load_examples(examples_csv_path)
@@ -307,10 +374,19 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                                 value=False
                             )
                             audio_prompt_input = gr.Audio(
-                                label="Audio Prompt (Optional - オプション)",
+                                label="Audio Prompt (Optional - オプション) [Max 10 seconds / 最大10秒]",
                                 sources=["upload", "microphone"],
                                 type="numpy"
                             )
+                            # Warning display for audio duration
+                            audio_warning_display = gr.HTML(value="", visible=True)
+                            
+                            # Audio validation on change
+                            audio_prompt_input.change(
+                                fn=on_audio_upload,
+                                inputs=[audio_prompt_input],
+                                outputs=[audio_prompt_input, audio_warning_display]
+                            )
                             
                             # Turbo mode event handler
                             def toggle_turbo(turbo_enabled):
@@ -332,7 +408,7 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                     status_output = gr.Textbox(label="Status", interactive=False)
                     audio_output = gr.Audio(label="Generated Speech", interactive=False, show_download_button=True)
                     
-            # Event handler
+            # Event handler - pass the warning display as a dummy input
             generate_button.click(
                 fn=run_generation_pipeline_client,
                 inputs=[
@@ -343,7 +419,8 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                     top_k_slider,
                     temperature_slider,
                     chained_longform_checkbox,
-                    seed_slider  # Add seed slider to inputs
+                    seed_slider,  # Add seed slider to inputs
+                    audio_warning_display  # Pass as dummy input
                 ],
                 outputs=[audio_output, status_output],
                 concurrency_limit=4  # Limit concurrent requests
@@ -434,10 +511,19 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                                 value=False
                             )
                             audio_prompt_input_dur = gr.Audio(
-                                label="Audio Prompt (Optional - オプション)",
+                                label="Audio Prompt (Optional - オプション) [Max 10 seconds / 最大10秒]",
                                 sources=["upload", "microphone"],
                                 type="numpy"
                             )
+                            # Warning display for audio duration
+                            audio_warning_display_dur = gr.HTML(value="", visible=True)
+                            
+                            # Audio validation on change
+                            audio_prompt_input_dur.change(
+                                fn=on_audio_upload,
+                                inputs=[audio_prompt_input_dur],
+                                outputs=[audio_prompt_input_dur, audio_warning_display_dur]
+                            )
                             
                             # Turbo mode event handler for duration tab
                             def toggle_turbo_dur(turbo_enabled):
@@ -459,7 +545,7 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                     status_output_dur = gr.Textbox(label="Status", interactive=False)
                     audio_output_dur = gr.Audio(label="Generated Speech", interactive=False, show_download_button=True)
                     
-            # Event handler for duration tab
+            # Event handler for duration tab - pass the warning display as a dummy input
             generate_button_dur.click(
                 fn=run_duration_generation_pipeline_client,
                 inputs=[
@@ -473,7 +559,8 @@ with gr.Blocks(theme="Respair/Shiki@9.1.0", css=css) as demo:
                     add_steps_slider,
                     use_duration_aware_checkbox,
                     chars_per_second_slider,
-                    seed_slider_dur  # Add seed slider to inputs
+                    seed_slider_dur,  # Add seed slider to inputs
+                    audio_warning_display_dur  # Pass as dummy input
                 ],
                 outputs=[audio_output_dur, status_output_dur],
                 concurrency_limit=4  # Limit concurrent requests