feat(app): minor update

.gitignore

@@ -10,6 +10,7 @@
 *.userosscache
 *.sln.docstates
 load.py
+button.py
 
 # User-specific files (MonoDevelop/Xamarin Studio)
 *.userprefs

app-full.py

@@ -27,6 +27,8 @@ limitation = os.getenv("SYSTEM") == "spaces"  # limit audio length in huggingfac
 def create_vc_fn(tgt_sr, net_g, vc, if_f0, file_index):
     def vc_fn(
         input_audio,
+        upload_audio,
+        upload_mode,
         f0_up_key,
         f0_method,
         index_rate,
@@ -43,20 +45,18 @@ def create_vc_fn(tgt_sr, net_g, vc, if_f0, file_index):
                 asyncio.run(edge_tts.Communicate(tts_text, "-".join(tts_voice.split('-')[:-1])).save("tts.mp3"))
                 audio, sr = librosa.load("tts.mp3", sr=16000, mono=True)
             else:
-                if config.files:
-                    audio, sr = librosa.load(input_audio, sr=16000, mono=True)
-                else:
+                if upload_mode:
                     if input_audio is None:
                         return "You need to upload an audio", None
-                    sampling_rate, audio = input_audio
+                    sampling_rate, audio = upload_audio
                     duration = audio.shape[0] / sampling_rate
-                    if duration > 20 and limitation:
-                        return "Please upload an audio file that is less than 20 seconds. If you need to generate a longer audio file, please use Colab.", None
                     audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32)
                     if len(audio.shape) > 1:
                         audio = librosa.to_mono(audio.transpose(1, 0))
                     if sampling_rate != 16000:
                         audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)
+                else:
+                    audio, sr = librosa.load(input_audio, sr=16000, mono=True)
             times = [0, 0, 0]
             f0_up_key = int(f0_up_key)
             audio_opt = vc.pipeline(
@@ -84,31 +84,31 @@ def create_vc_fn(tgt_sr, net_g, vc, if_f0, file_index):
 
 def cut_vocal_and_inst(yt_url):
     if yt_url != "":
-        if not os.path.exists("/content/youtube_audio"):
-            os.mkdir("/content/youtube_audio")
+        if not os.path.exists("youtube_audio"):
+            os.mkdir("youtube_audio")
         ydl_opts = {
             'format': 'bestaudio/best',
             'postprocessors': [{
                 'key': 'FFmpegExtractAudio',
                 'preferredcodec': 'wav',
             }],
-            "outtmpl": '/content/youtube_audio/audio',
+            "outtmpl": 'youtube_audio/audio',
         }
         with yt_dlp.YoutubeDL(ydl_opts) as ydl:
             ydl.download([yt_url])
-        yt_audio_path = "/content/youtube_audio/audio.wav"
+        yt_audio_path = "youtube_audio/audio.wav"
         command = f"demucs --two-stems=vocals {yt_audio_path}"
         result = subprocess.run(command.split(), stdout=subprocess.PIPE)
         print(result.stdout.decode())
-        return ("/content/rvc-models-new/separated/htdemucs/audio/vocals.wav", "/content/rvc-models-new/separated/htdemucs/audio/no_vocals.wav", yt_audio_path, "/content/rvc-models-new/separated/htdemucs/audio/vocals.wav")
+        return ("separated/htdemucs/audio/vocals.wav", "separated/htdemucs/audio/no_vocals.wav", yt_audio_path, "separated/htdemucs/audio/vocals.wav")
 
 def combine_vocal_and_inst(audio_data, audio_volume):
     print(audio_data)
-    if not os.path.exists("/content/result"):
-        os.mkdir("/content/result")
-    vocal_path = "/content/result/output.wav"
-    inst_path = "/content/rvc-models-new/separated/htdemucs/audio/no_vocals.wav"
-    output_path = "/content/result/combine.mp3"
+    if not os.path.exists("result"):
+        os.mkdir("result")
+    vocal_path = "result/output.wav"
+    inst_path = "separated/htdemucs/audio/no_vocals.wav"
+    output_path = "result/combine.mp3"
     with wave.open(vocal_path, "w") as wave_file:
         wave_file.setnchannels(1) 
         wave_file.setsampwidth(2)
@@ -132,19 +132,20 @@ def load_hubert():
         hubert_model = hubert_model.float()
     hubert_model.eval()
 
-def change_to_tts_mode(tts_mode):
+def change_to_tts_mode(tts_mode, upload_mode):
     if tts_mode:
-        return gr.Audio.update(visible=False), gr.Textbox.update(visible=True), gr.Dropdown.update(visible=True)
+        return gr.Textbox.update(visible=False), gr.Audio.update(visible=False), gr.Checkbox.update(visible=False), gr.Textbox.update(visible=True), gr.Dropdown.update(visible=True)
     else:
-        return gr.Audio.update(visible=True), gr.Textbox.update(visible=False), gr.Dropdown.update(visible=False)
+        if upload_mode:
+            return gr.Textbox.update(visible=False), gr.Audio.update(visible=True), gr.Checkbox.update(visible=True), gr.Textbox.update(visible=False), gr.Dropdown.update(visible=False)
+        else:
+            return gr.Textbox.update(visible=True), gr.Audio.update(visible=False), gr.Checkbox.update(visible=True), gr.Textbox.update(visible=False), gr.Dropdown.update(visible=False)
 
-'''
-def change_audio_to_path_mode(audio_mode):
-    if audio_mode:
-        return gr.Audio.update(visible=False), gr.Textbox.update(visible=True)
+def change_to_upload_mode(upload_mode):
+    if upload_mode:
+        return gr.Textbox().update(visible=False), gr.Audio().update(visible=True)
     else:
-        return gr.Audio.update(visible=True), gr.Textbox.update(visible=False)
-'''
+        return gr.Textbox().update(visible=True), gr.Audio().update(visible=False)
 
 if __name__ == '__main__':
     load_hubert()
@@ -215,7 +216,7 @@ if __name__ == '__main__':
             "# <center> RVC Models (Latest Update)\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> Please regenerate your model to latest RVC to fully applied this new rvc.\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)"
         )
@@ -231,19 +232,16 @@ if __name__ == '__main__':
                             '</div>'
                         )
                     with gr.Row():
-                        if config.files:
-                            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_audio_mode = gr.Checkbox(label="Use audio path", value=False)
-                            if config.files:
-                                vc_input = gr.Textbox(label="Input audio path")
-                            else:
-                                vc_input = gr.Audio(label="Input audio"+' (less than 20 seconds)' if limitation else '')
+                            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",
@@ -261,28 +259,27 @@ if __name__ == '__main__':
                             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_submit = gr.Button("Generate", variant="primary")
                             vc_output1 = gr.Textbox(label="Output Message")
                             vc_output2 = gr.Audio(label="Output Audio")
-                        if config.files:
-                            with gr.Column():
-                                vc_volume = gr.Slider(
-                                    minimum=0,
-                                    maximum=10,
-                                    label="Vocal volume",
-                                    value=5,
-                                    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_transpose, vc_f0method, vc_index_ratio, tts_mode, tts_text, tts_voice], [vc_output1, vc_output2])
-                tts_mode.change(change_to_tts_mode, [tts_mode], [vc_input, tts_text, tts_voice])
-                # vc_audio_input(change_audio_to_path_mode, [vc_audio_mode])
-                if config.files:
-                    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)
-        gr.Markdown('# <center>Changelog 2023.05.14')
-        gr.Markdown('- Remove nilou-jp due to outdated model')
-        gr.Markdown('- Added ganyu-jp from latest rvc train')
+                            vc_submit = gr.Button("Generate", variant="primary")
+                        with gr.Column():
+                            vc_volume = gr.Slider(
+                                minimum=0,
+                                maximum=10,
+                                label="Vocal volume",
+                                value=3,
+                                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])
+        gr.Markdown('# <center>Changelog 2023.05.15')
+        gr.Markdown('- Added support for direct upload to gradio')
+        gr.Markdown('- Added ayato-jp')
+        gr.Markdown('- Minor fix and adjustment')
         app.queue(concurrency_count=1, max_size=20, api_open=config.api).launch(share=config.colab)

app.py

@@ -166,7 +166,7 @@ if __name__ == '__main__':
             "# <center> RVC Models (Latest Update)\n"
             "## <center> The input audio should be clean and pure voice without background music.\n"
             "### <center> [Recommended to use google colab for more features](https://colab.research.google.com/drive/110kiMZTdP6Ri1lY9-NbQf17GVPPhHyeT?usp=sharing) \n"
-            "##### <center> Please regenerate your model to latest RVC to fully applied this new rvc.\n"
+            "#### <center> Please regenerate your model to latest RVC to fully applied this new rvc.\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)"
         )
@@ -207,7 +207,8 @@ if __name__ == '__main__':
                             vc_output2 = gr.Audio(label="Output Audio")
                 vc_submit.click(vc_fn, [vc_input, vc_transpose, vc_f0method, vc_index_ratio, tts_mode, tts_text, tts_voice], [vc_output1, vc_output2])
                 tts_mode.change(change_to_tts_mode, [tts_mode], [vc_input, tts_text, tts_voice])
-        gr.Markdown('# <center>Changelog 2023.05.14')
-        gr.Markdown('- Disable nilou-jp due to an outdated model')
-        gr.Markdown('- Added ganyu-jp from latest rvc train')
+        gr.Markdown('# <center>Changelog 2023.05.15')
+        gr.Markdown('- Added support for direct upload to gradio')
+        gr.Markdown('- Added ayato-jp')
+        gr.Markdown('- Minor fix and adjustment')
         app.queue(concurrency_count=1, max_size=20, api_open=config.api).launch(share=config.colab)

config.py

@@ -17,7 +17,6 @@ class Config:
             self.noparallel,
             self.noautoopen,
             self.api,
-            self.files,
             self.json
         ) = self.arg_parse()
         self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()
@@ -39,7 +38,6 @@ class Config:
             help="Do not open in browser automatically",
         )
         parser.add_argument('--api', action="store_true", default=False)
-        parser.add_argument("--files", action="store_true", default=False, help="load audio from path")
         parser.add_argument("--json", action="store_true", default=False, help="use model_info.json")
         cmd_opts = parser.parse_args()
 
@@ -52,7 +50,6 @@ class Config:
             cmd_opts.noparallel,
             cmd_opts.noautoopen,
             cmd_opts.api,
-            cmd_opts.files,
             cmd_opts.json
         )
 

weights/ayato-jp/added_IVF1304_Flat_nprobe_1.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fd600f40f3d8ef0c490eb66d73f25f89056d982edd4f467bda6b699d13e1fca
+size 53868475

weights/ayato-jp/ayato-jp.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6139fd4482942ab0075f08d5960d8b9b576945e6204ef2965dd3bb96ccccca4
+size 55027130

weights/model_info.json

@@ -14,5 +14,13 @@
         "cover": "cover.png",
 		"feature_retrieval_library": "added_IVF1636_Flat_nprobe_1.index",
 		"author":"ArkanDash"
+    },
+	"ayato-jp": {
+		"enable": true,
+		"name": "ayato-jp",
+        "title": "Genshin Impact - Kamisato Ayato",
+        "cover": "cover.png",
+		"feature_retrieval_library": "added_IVF1304_Flat_nprobe_1.index",
+		"author":"ArkanDash"
     }
 }