Rearranged a lot for the sake of client memory optimization.

app.py

@@ -406,6 +406,77 @@ if __name__ == '__main__':
     load_hubert()
     categories = load_model()
     voices = list(language_dict.keys())
+
+    #Gradio preloading
+    vc_audio_mode = gr.Dropdown(label="Input voice", choices=audio_mode, allow_custom_value=False, value="Edge-TTS")
+    # Input and Upload
+    vc_input = gr.Textbox(label="Input audio path", visible=False)
+    vc_upload = gr.Audio(label="Upload audio file", visible=False, interactive=True)
+    # Youtube
+    vc_download_audio = gr.Dropdown(label="Provider", choices=["Youtube"], allow_custom_value=False, visible=False, value="Youtube", info="Select provider (Default: Youtube)")
+    vc_link = gr.Textbox(label="Youtube URL", visible=False, info="Example: https://www.youtube.com/watch?v=Nc0sB1Bmf-A", placeholder="https://www.youtube.com/watch?v=...")
+    vc_split_model = gr.Dropdown(label="Splitter Model", choices=["htdemucs", "mdx_extra_q"], allow_custom_value=False, visible=False, value="htdemucs", info="Select the splitter model (Default: htdemucs)")
+    vc_split = gr.Button("Split Audio", variant="primary", visible=False)
+    vc_vocal_preview = gr.Audio(label="Vocal Preview", visible=False)
+    vc_inst_preview = gr.Audio(label="Instrumental Preview", visible=False)
+    vc_audio_preview = gr.Audio(label="Audio Preview", visible=False)
+    # TTS
+    tts_text = gr.Textbox(visible=True, label="TTS text", info="Text to speech input (There is a limit of 250 characters)", interactive=True)
+    tts_voice = gr.Dropdown(label="Edge-tts speaker", choices=voices, visible=True, allow_custom_value=False, value="English-Ana (Female)", interactive=True)
+    # Record Own
+    record_button = gr.Audio(source="microphone", label="Record your own audio", visible=False, interactive=True)
+    vc_transform0 = gr.Number(label="Transpose", value=0, info='Type "12" to change from male to female voice. Type "-12" to change female to male voice')
+    f0method0 = gr.Radio(
+        label="Pitch extraction algorithm",
+        info=f0method_info,
+        choices=f0method_mode,
+        value="pm",
+        interactive=True
+    )
+    index_rate1 = gr.Slider(
+        minimum=0,
+        maximum=1,
+        label="Retrieval feature ratio",
+        info="Accent control. Too high will usually sound too robotic. (Default: 0.4)",
+        value=0.4,
+        interactive=True,
+    )
+    filter_radius0 = gr.Slider(
+        minimum=0,
+        maximum=7,
+        label="Apply Median Filtering",
+        info="The value represents the filter radius and can reduce breathiness.",
+        value=1,
+        step=1,
+        interactive=True,
+    )
+    resample_sr0 = gr.Slider(
+        minimum=0,
+        maximum=48000,
+        label="Resample the output audio",
+        info="Resample the output audio in post-processing to the final sample rate. Set to 0 for no resampling.",
+        value=0,
+        step=1,
+        interactive=True,
+    )
+    rms_mix_rate0 = gr.Slider(
+        minimum=0,
+        maximum=1,
+        label="Volume Envelope",
+        info="Use the volume envelope of the input to replace or mix with the volume envelope of the output. The closer the ratio is to 1, the more the output envelope is used",
+        value=1,
+        interactive=True,
+    )
+    protect0 = gr.Slider(
+        minimum=0,
+        maximum=0.5,
+        label="Voice Protection",
+        info="Protect voiceless consonants and breath sounds to prevent artifacts such as tearing in electronic music. Set to 0.5 to disable. Decrease the value to increase protection, but it may reduce indexing accuracy",
+        value=0.23,
+        step=0.01,
+        interactive=True,
+    )
+
     with gr.Blocks(theme=gr.themes.Base()) as app:
         gr.Markdown(
             "# <center> VTuber RVC Models\n"
@@ -433,81 +504,6 @@ if __name__ == '__main__':
                                     (f'<img style="width:auto;height:300px;" src="file/{cover}">' if cover else "")+
                                     '</div>'
                                 )
-                            with gr.Row():
-                                with gr.Column():
-                                    vc_audio_mode = gr.Dropdown(label="Input voice", choices=audio_mode, allow_custom_value=False, value="Edge-TTS")
-                                    # Input and Upload
-                                    vc_input = gr.Textbox(label="Input audio path", visible=False)
-                                    vc_upload = gr.Audio(label="Upload audio file", visible=False, interactive=True)
-                                    # Youtube
-                                    vc_download_audio = gr.Dropdown(label="Provider", choices=["Youtube"], allow_custom_value=False, visible=False, value="Youtube", info="Select provider (Default: Youtube)")
-                                    vc_link = gr.Textbox(label="Youtube URL", visible=False, info="Example: https://www.youtube.com/watch?v=Nc0sB1Bmf-A", placeholder="https://www.youtube.com/watch?v=...")
-                                    vc_split_model = gr.Dropdown(label="Splitter Model", choices=["htdemucs", "mdx_extra_q"], allow_custom_value=False, visible=False, value="htdemucs", info="Select the splitter model (Default: htdemucs)")
-                                    vc_split = gr.Button("Split Audio", variant="primary", visible=False)
-                                    vc_vocal_preview = gr.Audio(label="Vocal Preview", visible=False)
-                                    vc_inst_preview = gr.Audio(label="Instrumental Preview", visible=False)
-                                    vc_audio_preview = gr.Audio(label="Audio Preview", visible=False)
-                                    # TTS
-                                    tts_text = gr.Textbox(visible=True, label="TTS text", info="Text to speech input (There is a limit of 250 characters)", interactive=True)
-                                    tts_voice = gr.Dropdown(label="Edge-tts speaker", choices=voices, visible=True, allow_custom_value=False, value="English-Ana (Female)", interactive=True)
-                                    # Record Own
-                                    record_button = gr.Audio(source="microphone", label="Record your own audio", visible=False, interactive=True)
-                                with gr.Column():
-                                    vc_transform0 = gr.Number(label="Transpose", value=0, info='Type "12" to change from male to female voice. Type "-12" to change female to male voice')
-                                    f0method0 = gr.Radio(
-                                        label="Pitch extraction algorithm",
-                                        info=f0method_info,
-                                        choices=f0method_mode,
-                                        value="rmvpe",
-                                        interactive=True
-                                    )
-                                    if model_index:
-                                        index_rate1 = gr.Slider(
-                                            minimum=0,
-                                            maximum=1,
-                                            label="Retrieval feature ratio",
-                                            info="Accent control. Too high will usually sound too robotic. (Default: 0.4)",
-                                            value=0.4,
-                                            interactive=True,
-                                        )
-                                    else:
-                                        index_rate1 = gr.Number(value=0.4, interactive=False, visible=False)
-                                    with gr.Accordion("Advanced Options", open=False):
-                                        filter_radius0 = gr.Slider(
-                                            minimum=0,
-                                            maximum=7,
-                                            label="Apply Median Filtering",
-                                            info="The value represents the filter radius and can reduce breathiness.",
-                                            value=1,
-                                            step=1,
-                                            interactive=True,
-                                        )
-                                        resample_sr0 = gr.Slider(
-                                            minimum=0,
-                                            maximum=48000,
-                                            label="Resample the output audio",
-                                            info="Resample the output audio in post-processing to the final sample rate. Set to 0 for no resampling.",
-                                            value=0,
-                                            step=1,
-                                            interactive=True,
-                                        )
-                                        rms_mix_rate0 = gr.Slider(
-                                            minimum=0,
-                                            maximum=1,
-                                            label="Volume Envelope",
-                                            info="Use the volume envelope of the input to replace or mix with the volume envelope of the output. The closer the ratio is to 1, the more the output envelope is used",
-                                            value=1,
-                                            interactive=True,
-                                        )
-                                        protect0 = gr.Slider(
-                                            minimum=0,
-                                            maximum=0.5,
-                                            label="Voice Protection",
-                                            info="Protect voiceless consonants and breath sounds to prevent artifacts such as tearing in electronic music. Set to 0.5 to disable. Decrease the value to increase protection, but it may reduce indexing accuracy",
-                                            value=0.23,
-                                            step=0.01,
-                                            interactive=True,
-                                        )
                                 with gr.Column():
                                     vc_log = gr.Textbox(label="Output Information", interactive=False)
                                     vc_output = gr.Audio(label="Output Audio", interactive=False)
@@ -516,6 +512,30 @@ if __name__ == '__main__':
                                     vc_mp = gr.Textbox(value=model_path, visible=False, interactive=False)
                                     vc_mi = gr.Textbox(value=model_index, visible=False, interactive=False)
                                     vc_convert = gr.Button("Convert", variant="primary")
+
+                                    vc_convert.click(
+                                        fn=infer,
+                                        inputs=[
+                                            vc_name,
+                                            vc_mp,
+                                            vc_mi,
+                                            vc_audio_mode,
+                                            vc_input, 
+                                            vc_upload,
+                                            tts_text,
+                                            tts_voice,
+                                            vc_transform0,
+                                            f0method0,
+                                            index_rate1,
+                                            filter_radius0,
+                                            resample_sr0,
+                                            rms_mix_rate0,
+                                            protect0,
+                                            record_button
+                                        ],
+                                        outputs=[vc_log, vc_output]
+                                    )
+                                    
                                     vc_volume = gr.Slider(
                                         minimum=0,
                                         maximum=10,
@@ -529,59 +549,65 @@ if __name__ == '__main__':
                                     vc_combined_output = gr.Audio(label="Output Combined Audio", visible=False)
                                     vc_combine =  gr.Button("Combine",variant="primary", visible=False)
 
-                        vc_convert.click(
-                            fn=infer,
-                            inputs=[
-                                vc_name,
-                                vc_mp,
-                                vc_mi,
-                                vc_audio_mode,
-                                vc_input, 
-                                vc_upload,
-                                tts_text,
-                                tts_voice,
-                                vc_transform0,
-                                f0method0,
-                                index_rate1,
-                                filter_radius0,
-                                resample_sr0,
-                                rms_mix_rate0,
-                                protect0,
-                                record_button
-                            ],
-                            outputs=[vc_log, vc_output]
-                        )
-                        vc_split.click(
-                            fn=cut_vocal_and_inst, 
-                            inputs=[vc_link, vc_download_audio, vc_split_model], 
-                            outputs=[vc_vocal_preview, vc_inst_preview, vc_audio_preview, vc_input]
-                        )
-                        vc_combine.click(
-                            fn=combine_vocal_and_inst,
-                            inputs=[vc_output, vc_volume, vc_split_model],
-                            outputs=[vc_combined_output]
-                        )
-                        vc_audio_mode.change(
-                            fn=change_audio_mode,
-                            inputs=[vc_audio_mode],
-                            outputs=[
-                                vc_input, 
-                                vc_upload,
-                                vc_download_audio,
-                                vc_link,
-                                vc_split_model,
-                                vc_split,
-                                vc_vocal_preview,
-                                vc_inst_preview,
-                                vc_audio_preview,
-                                vc_volume,
-                                vc_combined_output,
-                                vc_combine,
-                                tts_text,
-                                tts_voice,
-                                record_button
-                            ]
-                        )
+        with gr.Row():
+            with gr.Column():
+                vc_audio_mode.render()
+                vc_input.render()
+                vc_upload.render()
+                # Youtube
+                vc_download_audio.render()
+                vc_link.render()
+                vc_split_model.render()
+                vc_split.render()
+                vc_vocal_preview.render()
+                vc_inst_preview.render()
+                vc_audio_preview.render()
+                # TTS
+                tts_text.render()
+                tts_voice.render()
+                # Record Own
+                record_button.render()
+            with gr.Column():
+                vc_transform0.render()
+                f0method0.render()
+                index_rate1.render()
+                with gr.Accordion("Advanced Options", open=False):
+                    filter_radius0.render()
+                    resample_sr0.render()
+                    rms_mix_rate0.render()
+                    protect0.render() 
+
+        vc_split.click(
+            fn=cut_vocal_and_inst, 
+            inputs=[vc_link, vc_download_audio, vc_split_model], 
+            outputs=[vc_vocal_preview, vc_inst_preview, vc_audio_preview, vc_input]
+        )
+        vc_combine.click(
+            fn=combine_vocal_and_inst,
+            inputs=[vc_output, vc_volume, vc_split_model],
+            outputs=[vc_combined_output]
+        )
+        vc_audio_mode.change(
+            fn=change_audio_mode,
+            inputs=[vc_audio_mode],
+            outputs=[
+                vc_input, 
+                vc_upload,
+                vc_download_audio,
+                vc_link,
+                vc_split_model,
+                vc_split,
+                vc_vocal_preview,
+                vc_inst_preview,
+                vc_audio_preview,
+                vc_volume,
+                vc_combined_output,
+                vc_combine,
+                tts_text,
+                tts_voice,
+                record_button
+            ]
+        )
         gr.Markdown(
             "## <center>Credit to:\n"
             "#### <center>Original devs:\n"