Skip to content

Commit 240c81c

Browse files
TheNeodevTheNeodev
committed
Created using Colab
1 parent 24a7a87 commit 240c81c

File tree

1 file changed

+113
-62
lines changed

1 file changed

+113
-62
lines changed

‎Advanced-RVC.ipynb

Lines changed: 113 additions & 62 deletions
Original file line numberDiff line numberDiff line change
@@ -80,75 +80,126 @@
8080
"!python ./models.py"
8181
]
8282
},
83-
{
84-
"cell_type": "code",
85-
"source": [
86-
"#@title youtube downloader for separation\n",
87-
"import yt_dlp\n",
88-
"import os\n",
89-
"\n",
90-
"def download_wav(video_url, output_dir):\n",
91-
" ydl_opts = {\n",
92-
" 'format': 'bestaudio/best',\n",
93-
" 'outtmpl': os.path.join(output_dir, '%(title)s.%(ext)s'),\n",
94-
" 'noplaylist': True, # Download only a single video, not a playlist\n",
95-
" 'extract_audio': True,\n",
96-
" 'audio_format': 'wav', # Ensure WAV format\n",
97-
" 'postprocessors': [{\n",
98-
" 'key': 'FFmpegExtractAudio',\n",
99-
" 'preferredcodec': 'wav',\n",
100-
" }]\n",
101-
" }\n",
102-
"\n",
103-
" try:\n",
104-
" with yt_dlp.YoutubeDL(ydl_opts) as ydl:\n",
105-
" ydl.download([video_url])\n",
106-
" print(f\"Download completed. Audio saved in {output_dir}\")\n",
107-
" except yt_dlp.utils.DownloadError as e:\n",
108-
" print(f\"An error occurred: {e}\")\n",
109-
"\n",
110-
"# Example usage\n",
111-
"video_url = 'Video_URL' #@param {type:\"string\"}\n",
112-
" # Replace with your YouTube video URL\n",
113-
"output_directory = '/content/yt_dlp' #@param {type:\"string\"}\n",
114-
" # customize your output directory\n",
115-
"\n",
116-
"# Create the output directory if it doesn't exist.\n",
117-
"os.makedirs(output_directory, exist_ok=True)\n",
118-
"\n",
119-
"download_wav(video_url, output_directory)"
120-
],
121-
"metadata": {
122-
"cellView": "form",
123-
"id": "hDYUwhTXJI3_"
124-
},
125-
"execution_count": null,
126-
"outputs": []
127-
},
12883
{
12984
"cell_type": "code",
13085
"source": [
13186
"#@title audio separator for infernece\n",
13287
"\n",
13388
"\n",
13489
"\n",
135-
"from audio_separator.separator import Separator\n",
136-
"\n",
137-
"aud_input = \"Input_aud\" #@param {type:\"string\"}\n",
138-
"\n",
139-
"output_dir = \"/content/separated\" #@param {type:\"string\"}\n",
140-
"\n",
141-
"# Initialize the Separator class (with optional configuration properties, below)\n",
142-
"separator = Separator(output_dir=output_dir, output_format=\"wav\")\n",
143-
"\n",
144-
"separator.load_model()\n",
90+
"import os\n",
91+
"import glob\n",
92+
"import yt_dlp\n",
14593
"\n",
146-
"output_names = {\n",
147-
" \"Vocals\": \"vocals_output\",\n",
148-
" \"Instrumental\": \"instrumental_output\",\n",
149-
"}\n",
150-
"output_files = separator.separate(aud_input, output_names)\n",
151-
"print(f\"Separation complete!\")"
94+
"def downloader(url):\n",
95+
" ydl_opts = {\n",
96+
" 'format': 'bestaudio/best',\n",
97+
" 'postprocessors': [{\n",
98+
" 'key': 'FFmpegExtractAudio',\n",
99+
" 'preferredcodec': 'wav',\n",
100+
" 'preferredquality': '192',\n",
101+
" }],\n",
102+
" 'outtmpl': os.path.join(f'{path}/temp', '%(title)s.%(ext)s'),\n",
103+
" }\n",
104+
"\n",
105+
" with yt_dlp.YoutubeDL(ydl_opts) as ydl:\n",
106+
" ydl.download([url])\n",
107+
"\n",
108+
"def checker(url):\n",
109+
" return \"http\" in url\n",
110+
"\n",
111+
"def uvr_cli(audio_input, output_folder, model, output_format, segment_size, overlap, batch_size, override_model_segment_size, use_autocast, extensions):\n",
112+
" found_files = []\n",
113+
"\n",
114+
" dictmodel = {\n",
115+
" 'BS-Roformer-Viperx-1297': 'model_bs_roformer_ep_317_sdr_12.9755.ckpt',\n",
116+
" 'BS-Roformer-Viperx-1296': 'model_bs_roformer_ep_368_sdr_12.9628.ckpt',\n",
117+
" 'BS-Roformer-Viperx-1053': 'model_bs_roformer_ep_937_sdr_10.5309.ckpt',\n",
118+
" 'Mel-Roformer-Viperx-1143': 'model_mel_band_roformer_ep_3005_sdr_11.4360.ckpt',\n",
119+
" 'BS-Roformer-De-Reverb': 'deverb_bs_roformer_8_384dim_10depth.ckpt',\n",
120+
" 'Mel-Roformer-Crowd-Aufr33-Viperx': 'mel_band_roformer_crowd_aufr33_viperx_sdr_8.7144.ckpt',\n",
121+
" 'Mel-Roformer-Denoise-Aufr33': 'denoise_mel_band_roformer_aufr33_sdr_27.9959.ckpt',\n",
122+
" 'Mel-Roformer-Denoise-Aufr33-Aggr' : 'denoise_mel_band_roformer_aufr33_aggr_sdr_27.9768.ckpt',\n",
123+
" 'Mel-Roformer-Karaoke-Aufr33-Viperx': 'mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt',\n",
124+
" 'MelBand Roformer | Vocals by Kimberley Jensen' : 'vocals_mel_band_roformer.ckpt',\n",
125+
" 'MelBand Roformer Kim | FT by unwa' : 'mel_band_roformer_kim_ft_unwa.ckpt',\n",
126+
" 'MelBand Roformer Kim | Inst V1 by Unwa' : 'melband_roformer_inst_v1.ckpt',\n",
127+
" 'MelBand Roformer Kim | Inst V1 (E) by Unwa' : 'melband_roformer_inst_v1e.ckpt',\n",
128+
" 'MelBand Roformer Kim | Inst V2 by Unwa' : 'melband_roformer_inst_v2.ckpt',\n",
129+
" 'MelBand Roformer Kim | InstVoc Duality V1 by Unwa' : 'melband_roformer_instvoc_duality_v1.ckpt',\n",
130+
" 'MelBand Roformer Kim | InstVoc Duality V2 by Unwa' : 'melband_roformer_instvox_duality_v2.ckpt',\n",
131+
" 'MelBand Roformer | De-Reverb by anvuew' : 'dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt',\n",
132+
" 'MelBand Roformer | De-Reverb Less Aggressive by anvuew' : 'dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt',\n",
133+
" 'MelBand Roformer | De-Reverb-Echo by Sucial' : 'dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt',\n",
134+
" 'MelBand Roformer | De-Reverb-Echo V2 by Sucial' : 'dereverb-echo_mel_band_roformer_sdr_13.4843_v2.ckpt',\n",
135+
" 'MelBand Roformer Kim | SYHFT by SYH99999' : 'MelBandRoformerSYHFT.ckpt',\n",
136+
" 'MelBand Roformer Kim | SYHFT V2 by SYH99999' : 'MelBandRoformerSYHFTV2.ckpt',\n",
137+
" 'MelBand Roformer Kim | SYHFT V2.5 by SYH99999' : 'MelBandRoformerSYHFTV2.5.ckpt',\n",
138+
" 'MelBand Roformer Kim | SYHFT V3 by SYH99999' : 'MelBandRoformerSYHFTV3Epsilon.ckpt',\n",
139+
" 'MelBand Roformer Kim | Big SYHFT V1 by SYH99999' : 'MelBandRoformerBigSYHFTV1.ckpt',\n",
140+
" 'MelBand Roformer Kim | Big Beta 4 FT by unwa' : 'melband_roformer_big_beta4.ckpt',\n",
141+
" 'MelBand Roformer Kim | Big Beta 5e FT by unwa' : 'melband_roformer_big_beta5e.ckpt',\n",
142+
" 'BS Roformer | Chorus Male-Female by Sucial' : 'model_chorus_bs_roformer_ep_267_sdr_24.1275.ckpt',\n",
143+
" 'MelBand Roformer | Aspiration by Sucial' : 'aspiration_mel_band_roformer_sdr_18.9845.ckpt',\n",
144+
" 'MelBand Roformer | Aspiration Less Aggressive by Sucial' : 'aspiration_mel_band_roformer_less_aggr_sdr_18.1201.ckpt',\n",
145+
" 'MelBand Roformer | Bleed Suppressor V1 by unwa-97chris' : 'mel_band_roformer_bleed_suppressor_v1.ckpt'\n",
146+
" }\n",
147+
" roformer_model = dictmodel[model]\n",
148+
"\n",
149+
" if checker(audio_input):\n",
150+
" downloader(audio_input)\n",
151+
" audio_input = f\"{path}/temp\"\n",
152+
"\n",
153+
" for audio_files in os.listdir(audio_input):\n",
154+
" if audio_files.endswith(extensions):\n",
155+
" found_files.append(audio_files)\n",
156+
"\n",
157+
" total_files = len(found_files)\n",
158+
"\n",
159+
" if total_files == 0:\n",
160+
" print(\"No valid audio files found.\")\n",
161+
" else:\n",
162+
" print(f\"{total_files} audio files found\")\n",
163+
"\n",
164+
" found_files.sort()\n",
165+
"\n",
166+
" for audio_files in found_files:\n",
167+
" file_path = os.path.join(audio_input, audio_files)\n",
168+
" prompt = f'audio-separator \"{file_path}\" --model_filename {roformer_model} --output_dir={output_folder} --output_format={output_format} --mdxc_segment_size={segment_size} --mdxc_overlap={overlap} --mdxc_batch_size={batch_size} --model_file_dir=./models'\n",
169+
" if override_model_segment_size:\n",
170+
" prompt += \" --mdxc_override_model_segment_size\"\n",
171+
" if use_autocast:\n",
172+
" prompt += \" --use_autocast\"\n",
173+
" !$prompt\n",
174+
"\n",
175+
" if audio_input == f\"{path}/temp\":\n",
176+
" temp_files = glob.glob(f\"{path}/temp/*\")\n",
177+
" for file in temp_files:\n",
178+
" os.remove(file)\n",
179+
"\n",
180+
"#@markdown Input path for audio files or link:\n",
181+
"audio_input = \"\" #@param {type:\"string\"}\n",
182+
"#@markdown You can paste the link to the video/audio from many sites, check the complete list [here](https://github.com/yt-dlp/yt-dlp/blob/master/supportedsites.md)\n",
183+
"\n",
184+
"#@markdown Output path for audio files:\n",
185+
"output_folder = \"/content/Vocales\" #@param {type:\"string\"}\n",
186+
"#@markdown Select the model:\n",
187+
"model = \"BS-Roformer-Viperx-1297\" #@param [\"BS-Roformer-Viperx-1297\", \"BS-Roformer-Viperx-1296\", \"BS-Roformer-Viperx-1053\", \"Mel-Roformer-Viperx-1143\", \"BS-Roformer-De-Reverb\", \"Mel-Roformer-Crowd-Aufr33-Viperx\", \"Mel-Roformer-Denoise-Aufr33\", \"Mel-Roformer-Denoise-Aufr33-Aggr\", \"Mel-Roformer-Karaoke-Aufr33-Viperx\", \"MelBand Roformer | Vocals by Kimberley Jensen\", \"MelBand Roformer Kim | FT by unwa\", \"MelBand Roformer Kim | Inst V1 by Unwa\", \"MelBand Roformer Kim | Inst V1 (E) by Unwa\", \"MelBand Roformer Kim | Inst V2 by Unwa\", \"MelBand Roformer Kim | InstVoc Duality V1 by Unwa\", \"MelBand Roformer Kim | InstVoc Duality V2 by Unwa\", \"MelBand Roformer | De-Reverb by anvuew\", \"MelBand Roformer | De-Reverb Less Aggressive by anvuew\", \"MelBand Roformer | De-Reverb-Echo by Sucial\", \"MelBand Roformer | De-Reverb-Echo V2 by Sucial\", \"MelBand Roformer Kim | SYHFT by SYH99999\", \"MelBand Roformer Kim | SYHFT V2 by SYH99999\", \"MelBand Roformer Kim | SYHFT V2.5 by SYH99999\", \"MelBand Roformer Kim | SYHFT V3 by SYH99999\", \"MelBand Roformer Kim | Big SYHFT V1 by SYH99999\", \"MelBand Roformer Kim | Big Beta 4 FT by unwa\", \"MelBand Roformer Kim | Big Beta 5e FT by unwa\", \"BS Roformer | Chorus Male-Female by Sucial\", \"MelBand Roformer | Aspiration by Sucial\", \"MelBand Roformer | Aspiration Less Aggressive by Sucial\", \"MelBand Roformer | Bleed Suppressor V1 by unwa-97chris\"]\n",
188+
"#@markdown Select the output format:\n",
189+
"output_format = \"wav\" #@param [\"wav\", \"flac\", \"mp3\", \"ogg\", \"opus\", \"m4a\", \"aiff\", \"ac3\"]\n",
190+
"#@markdown Larger consumes more resources, but may give better results.\n",
191+
"segment_size = 256 #@param {type:\"slider\", min:32, max:4000, step:32}\n",
192+
"#@markdown Amount of overlap between prediction windows.\n",
193+
"overlap = 8 #@param {type:\"slider\", min:2, max:10, step:1}\n",
194+
"#@markdown Larger consumes more RAM but may process slightly faster.\n",
195+
"batch_size = 1 #@param {type:\"slider\", min:1, max:16, step:1}\n",
196+
"#@markdown Override model default segment size instead of using the model default value.\n",
197+
"override_model_segment_size = False #@param {type:\"boolean\"}\n",
198+
"#@markdown Flag to use PyTorch autocast for faster inference. Do not use for CPU inference.\n",
199+
"use_autocast = True #@param {type:\"boolean\"}\n",
200+
"extensions = (\".wav\", \".flac\", \".mp3\", \".ogg\", \".opus\", \".m4a\", \".aiff\", \".ac3\")\n",
201+
"\n",
202+
"uvr_cli(audio_input, output_folder, model, output_format, segment_size, overlap, batch_size, override_model_segment_size, use_autocast, extensions)"
152203
],
153204
"metadata": {
154205
"cellView": "form",
@@ -227,7 +278,7 @@
227278
"\n",
228279
"#@title ## Inference\n",
229280
"import os\n",
230-
"main_dir = \"/content/Harmonify\"\n",
281+
"main_dir = os.getcwd()\n",
231282
"os.chdir(main_dir)\n",
232283
"from lib.infer import infer_audio\n",
233284
"from google.colab import files\n",

0 commit comments

Comments
 (0)