feat(app): added category and yelan-jp

app-full.py

@@ -150,10 +150,21 @@ def change_to_upload_mode(upload_mode):
 if __name__ == '__main__':
     load_hubert()
     models = []
+    categories = []
     tts_voice_list = asyncio.get_event_loop().run_until_complete(edge_tts.list_voices())
     voices = [f"{v['ShortName']}-{v['Gender']}" for v in tts_voice_list]
-    if config.json:
-        with open("weights/model_info.json", "r", encoding="utf-8") as f:
+    with open("weights/folder_info.json", "r", encoding="utf-8") as f:
+        folder_info = json.load(f)
+    for name, info in folder_info.items():
+        if not info['enable']:
+            continue
+        title = name
+        folder = info['folder_path']
+        cover = f"{folder}/{info['cover']}"
+        markdown = info['markdown']
+        catergories.append([title, folder, cover, markdown])
+    for (title, folder, cover, markdown) in categories:
+        with open(f"weights/{folder}/model_info.json", "r", encoding="utf-8") as f:
             models_info = json.load(f)
         for name, info in models_info.items():
             if not info['enable']:
@@ -178,104 +189,82 @@ if __name__ == '__main__':
             else:
                 net_g = net_g.float()
             vc = VC(tgt_sr, config)
-            models.append((name, title, author, cover, create_vc_fn(tgt_sr, net_g, vc, if_f0, index)))
-    else:
-        folder_path = "weights"
-        for name in os.listdir(folder_path):
-            print("check folder: " + name)
-            if name.startswith("."): break
-            cover_path = glob.glob(f"{folder_path}/{name}/*.png") + glob.glob(f"{folder_path}/{name}/*.jpg")
-            index_path = glob.glob(f"{folder_path}/{name}/*.index")
-            checkpoint_path = glob.glob(f"{folder_path}/{name}/*.pth")
-            title = name
-            author = ""
-            if cover_path:
-                cover = cover_path[0]
-            else:
-                cover = ""
-            index = index_path[0]
-            cpt = torch.load(checkpoint_path[0], map_location="cpu")
-            tgt_sr = cpt["config"][-1]
-            cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]  # n_spk
-            if_f0 = cpt.get("f0", 1)
-            if if_f0 == 1:
-                net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=config.is_half)
-            else:
-                net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
-            del net_g.enc_q
-            print(net_g.load_state_dict(cpt["weight"], strict=False))  # 不加这一行清不干净, 真奇葩
-            net_g.eval().to(config.device)
-            if config.is_half:
-                net_g = net_g.half()
-            else:
-                net_g = net_g.float()
-            vc = VC(tgt_sr, config)
+            print(f"Model loaded: {name}")
             models.append((name, title, author, cover, create_vc_fn(tgt_sr, net_g, vc, if_f0, index)))
     with gr.Blocks() as app:
         gr.Markdown(
             "# <center> RVC Models [(Latest Update)](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases/tag/20230428updated)\n"
             "## <center> The input audio should be clean and pure voice without background music.\n"
-            "### <center> More feature will be added soon... \n"
-            "#### <center> Please regenerate your model to latest RVC to fully applied this new rvc.\n"
+            "### <center> This project was inspired by [zomehwh](https://huggingface.co/spaces/zomehwh/rvc-models) and [ardha27](https://huggingface.co/spaces/ardha27/rvc-models)\n"
             "[![image](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/110kiMZTdP6Ri1lY9-NbQf17GVPPhHyeT?usp=sharing)\n\n"
             "[![Original Repo](https://badgen.net/badge/icon/github?icon=github&label=Original%20Repo)](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI)"
         )
-        with gr.Tabs():
-            for (name, title, author, cover, vc_fn) in models:
-                with gr.TabItem(name):
-                    with gr.Row():
-                        gr.Markdown(
-                            '<div align="center">'
-                            f'<div>{title}</div>\n'+
-                            (f'<div>Model author: {author}</div>' if author else "")+
-                            (f'<img style="width:auto;height:300px;" src="file/{cover}">' if cover else "")+
-                            '</div>'
-                        )
-                    with gr.Row():
-                        with gr.Column():
-                            vc_youtube = gr.Textbox(label="Youtube URL")
-                            vc_convert = gr.Button("Convert", variant="primary")
-                            vc_vocal_preview = gr.Audio(label="Vocal Preview")
-                            vc_inst_preview = gr.Audio(label="Instrumental Preview")
-                            vc_audio_preview = gr.Audio(label="Audio Preview")
-                        with gr.Column():
-                            vc_input = gr.Textbox(label="Input audio path")
-                            vc_upload = gr.Audio(label="Upload audio file", visible=False, interactive=True)
-                            upload_mode = gr.Checkbox(label="Upload mode", value=False)
-                            vc_transpose = gr.Number(label="Transpose", value=0)
-                            vc_f0method = gr.Radio(
-                                label="Pitch extraction algorithm, PM is fast but Harvest is better for low frequencies",
-                                choices=["pm", "harvest"],
-                                value="pm",
-                                interactive=True,
-                            )
-                            vc_index_ratio = gr.Slider(
-                                minimum=0,
-                                maximum=1,
-                                label="Retrieval feature ratio",
-                                value=0.6,
-                                interactive=True,
-                            )
-                            tts_mode = gr.Checkbox(label="tts (use edge-tts as input)", value=False)
-                            tts_text = gr.Textbox(visible=False,label="TTS text (100 words limitation)" if limitation else "TTS text")
-                            tts_voice = gr.Dropdown(label="Edge-tts speaker", choices=voices, visible=False, allow_custom_value=False, value="en-US-AnaNeural-Female")
-                            vc_output1 = gr.Textbox(label="Output Message")
-                            vc_output2 = gr.Audio(label="Output Audio")
-                            vc_submit = gr.Button("Generate", variant="primary")
-                        with gr.Column():
-                            vc_volume = gr.Slider(
-                                minimum=0,
-                                maximum=10,
-                                label="Vocal volume",
-                                value=4,
-                                interactive=True,
-                                step=1
-                            )
-                            vc_outputCombine = gr.Audio(label="Output Combined Audio")
-                            vc_combine =  gr.Button("Combine",variant="primary")
-                vc_submit.click(vc_fn, [vc_input, vc_upload, upload_mode, vc_transpose, vc_f0method, vc_index_ratio, tts_mode, tts_text, tts_voice], [vc_output1, vc_output2])
-                vc_convert.click(cut_vocal_and_inst, vc_youtube, [vc_vocal_preview, vc_inst_preview, vc_audio_preview, vc_input])
-                vc_combine.click(combine_vocal_and_inst, [vc_output2, vc_volume], vc_outputCombine)
-                tts_mode.change(change_to_tts_mode, [tts_mode, upload_mode], [vc_input, vc_upload, upload_mode, tts_text, tts_voice])
-                upload_mode.change(change_to_upload_mode, [upload_mode], [vc_input, vc_upload])
+        for (title, folder, cover, markdown) in categories:
+            gr.Markdown(
+                '<div align="center">'
+                (f'<img style="width:auto;height:500px;" src="file/{cover}">' if cover else "")+
+                '<div>'
+            )
+            with gr.TabItem(title):
+                with gr.Tabs():
+                    if not models == True:
+                        gr.Markdown("# <center> No Model Loaded.")
+                        return gr.Markdown("## <center> Please added the model or fix your model path.")
+                    for (name, title, author, cover, vc_fn) in models:
+                        with gr.TabItem(name):
+                            with gr.Row():
+                                gr.Markdown(
+                                    '<div align="center">'
+                                    f'<div>{title}</div>\n'+
+                                    (f'<div>Model author: {author}</div>' if author else "")+
+                                    (f'<img style="width:auto;height:300px;" src="file/{cover}">' if cover else "")+
+                                    '</div>'
+                                )
+                            with gr.Row():
+                                with gr.Column():
+                                    vc_youtube = gr.Textbox(label="Youtube URL")
+                                    vc_convert = gr.Button("Convert", variant="primary")
+                                    vc_vocal_preview = gr.Audio(label="Vocal Preview")
+                                    vc_inst_preview = gr.Audio(label="Instrumental Preview")
+                                    vc_audio_preview = gr.Audio(label="Audio Preview")
+                                with gr.Column():
+                                    vc_input = gr.Textbox(label="Input audio path")
+                                    vc_upload = gr.Audio(label="Upload audio file", visible=False, interactive=True)
+                                    upload_mode = gr.Checkbox(label="Upload mode", value=False)
+                                    vc_transpose = gr.Number(label="Transpose", value=0)
+                                    vc_f0method = gr.Radio(
+                                        label="Pitch extraction algorithm, PM is fast but Harvest is better for low frequencies",
+                                        choices=["pm", "harvest"],
+                                        value="pm",
+                                        interactive=True,
+                                    )
+                                    vc_index_ratio = gr.Slider(
+                                        minimum=0,
+                                        maximum=1,
+                                        label="Retrieval feature ratio",
+                                        value=0.6,
+                                        interactive=True,
+                                    )
+                                    tts_mode = gr.Checkbox(label="tts (use edge-tts as input)", value=False)
+                                    tts_text = gr.Textbox(visible=False,label="TTS text (100 words limitation)" if limitation else "TTS text")
+                                    tts_voice = gr.Dropdown(label="Edge-tts speaker", choices=voices, visible=False, allow_custom_value=False, value="en-US-AnaNeural-Female")
+                                    vc_output1 = gr.Textbox(label="Output Message")
+                                    vc_output2 = gr.Audio(label="Output Audio")
+                                    vc_submit = gr.Button("Generate", variant="primary")
+                                with gr.Column():
+                                    vc_volume = gr.Slider(
+                                        minimum=0,
+                                        maximum=10,
+                                        label="Vocal volume",
+                                        value=4,
+                                        interactive=True,
+                                        step=1
+                                    )
+                                    vc_outputCombine = gr.Audio(label="Output Combined Audio")
+                                    vc_combine =  gr.Button("Combine",variant="primary")
+                        vc_submit.click(vc_fn, [vc_input, vc_upload, upload_mode, vc_transpose, vc_f0method, vc_index_ratio, tts_mode, tts_text, tts_voice], [vc_output1, vc_output2])
+                        vc_convert.click(cut_vocal_and_inst, vc_youtube, [vc_vocal_preview, vc_inst_preview, vc_audio_preview, vc_input])
+                        vc_combine.click(combine_vocal_and_inst, [vc_output2, vc_volume], vc_outputCombine)
+                        tts_mode.change(change_to_tts_mode, [tts_mode, upload_mode], [vc_input, vc_upload, upload_mode, tts_text, tts_voice])
+                        upload_mode.change(change_to_upload_mode, [upload_mode], [vc_input, vc_upload])
         app.queue(concurrency_count=1, max_size=20, api_open=config.api).launch(share=config.colab)

config.py

@@ -16,8 +16,7 @@ class Config:
             self.colab,
             self.noparallel,
             self.noautoopen,
-            self.api,
-            self.json
+            self.api
         ) = self.arg_parse()
         self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()
 
@@ -38,7 +37,6 @@ class Config:
             help="Do not open in browser automatically",
         )
         parser.add_argument('--api', action="store_true", default=False)
-        parser.add_argument("--json", action="store_true", default=False, help="use model_info.json")
         cmd_opts = parser.parse_args()
 
         cmd_opts.port = cmd_opts.port if 0 <= cmd_opts.port <= 65535 else 7865
@@ -50,7 +48,6 @@ class Config:
             cmd_opts.noparallel,
             cmd_opts.noautoopen,
             cmd_opts.api,
-            cmd_opts.json
         )
 
     def device_config(self) -> tuple:

weights/folder_info.json

@@ -0,0 +1,9 @@
+{
+    "genshin-impact":{
+        "enable": true,
+        "title": "Genshin Impact",
+        "folder_path": "genshin-impact",
+        "cover": "cover.png",
+        "markdown": ""
+    }
+}

weights/{model_info.json → genshin-impact/model_info.json}

@@ -47,6 +47,14 @@
 		"feature_retrieval_library": "added_IVF1723_Flat_nprobe_1.index",
 		"author":"ArkanDash"
 	},
+	"yelan-jp": {
+		"enable": true,
+		"name": "yelan-jp",
+        "title": "Genshin Impact - Yelan",
+        "cover": "cover.png",
+		"feature_retrieval_library": "added_IVF2051_Flat_nprobe_1.index",
+		"author":"ArkanDash"
+	},
 	"yoimiya-jp": {
 		"enable": true,
 		"name": "yoimiya-jp",

weights/genshin-impact/yelan-jp/added_IVF2051_Flat_nprobe_1.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba725109af57b0c32075582e17571e59ee3a3c33b34b00e13481bd15c0bf0920
+size 84694315

weights/genshin-impact/yelan-jp/yelan-jp.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d62ff825db07022e7d380d13c827f27730c99f6d029e8fb8b1d7cd90018d3c49
+size 55027130