diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000000000000000000000000000000000000..634fad90560fb545bdf1b6e65751bda8b13ac637 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,47 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +Resources/ffmpeg.exe filter=lfs diff=lfs merge=lfs -text +Resources/ffprobe.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.0.9.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.1.0.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.1.1.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.5.0.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.2.0.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.2.1.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.2.2.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.2.3.exe filter=lfs diff=lfs merge=lfs -text +Compiled/ApplioV3.2.4.exe filter=lfs diff=lfs merge=lfs -text +Resources/formant/stftpitchshift filter=lfs diff=lfs merge=lfs -text diff --git a/Compiled/.gitignore b/Compiled/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Compiled/Linux/.gitignore b/Compiled/Linux/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Compiled/Linux/ApplioV3.3.0.zip b/Compiled/Linux/ApplioV3.3.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..bbe7255a3cee23da234250ee0174d864f11c0c4e --- /dev/null +++ b/Compiled/Linux/ApplioV3.3.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a10298f10f85e8c49fa8d7490053fde5af3ce76446eb4d812cdfd3db9715c71 +size 9575763753 diff --git a/Compiled/Windows/.gitignore b/Compiled/Windows/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Compiled/Windows/ApplioV3.2.8-bugfix.zip b/Compiled/Windows/ApplioV3.2.8-bugfix.zip new file mode 100644 index 0000000000000000000000000000000000000000..9100fdecefc449177622b67c2dd590c14daea93c --- /dev/null +++ b/Compiled/Windows/ApplioV3.2.8-bugfix.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42c68274095cc9d11930dd4a783acea7a9de641753ea1033431fbb07a88ef74b +size 4220255228 diff --git a/Compiled/Windows/ApplioV3.2.9.zip b/Compiled/Windows/ApplioV3.2.9.zip new file mode 100644 index 0000000000000000000000000000000000000000..9871fb99e2bc6e8b9c3c666628e174e3f423cb66 --- /dev/null +++ b/Compiled/Windows/ApplioV3.2.9.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87e6329870f36a31106913a627a1602a1abd84085884d875f83bd53f05b82941 +size 4236244130 diff --git a/Compiled/Windows/ApplioV3.3.0.zip b/Compiled/Windows/ApplioV3.3.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..34f62ad1a7b5c251e73601586e6af0c1a66254cd --- /dev/null +++ b/Compiled/Windows/ApplioV3.3.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:641bddfded4e21b735397572420b34d09d1905c72939a9d629ce981f5fc68ea2 +size 4873873650 diff --git a/Compiled/Windows/ApplioV3.3.1.zip b/Compiled/Windows/ApplioV3.3.1.zip new file mode 100644 index 0000000000000000000000000000000000000000..13d1a5a270b9949174ca1664e15d236da522eb71 --- /dev/null +++ b/Compiled/Windows/ApplioV3.3.1.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af3742262e0ea6724eeca188ad95f9910eccd658321ee2f8f2a88b51df6b4f50 +size 4879206120 diff --git a/Compiled/Windows/ApplioV3.4.0.zip b/Compiled/Windows/ApplioV3.4.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..4d51bf21a2eecffe011d73f04cd012978d323b12 --- /dev/null +++ b/Compiled/Windows/ApplioV3.4.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:071fb5ec31e12c7afc90aa3a8b1af4842840b44a7b14ccb2fd2b60e9e5268344 +size 4879293026 diff --git a/Compiled/Windows/ApplioV3.5.0.zip b/Compiled/Windows/ApplioV3.5.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..3105200e13ac063cf713a126d489d8521f627ed7 --- /dev/null +++ b/Compiled/Windows/ApplioV3.5.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dbe741466989905890d1e046909a6f8d26feaa0244c5defb43e8e116cf4cb6df +size 4899282098 diff --git a/Compiled/Windows/ApplioV3.5.1.zip b/Compiled/Windows/ApplioV3.5.1.zip new file mode 100644 index 0000000000000000000000000000000000000000..ae997e1701d0928db3280e41d68c54083daa227b --- /dev/null +++ b/Compiled/Windows/ApplioV3.5.1.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d8ac88f58aeb265e8a6651085479106410c3602cbe3f64918bc63f64ad768ec +size 4862979406 diff --git a/Compiled/Windows/ApplioV3.6.0.zip b/Compiled/Windows/ApplioV3.6.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..88bb95fec15569c98eddb437420d1c8285816653 --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dc30842d9dd28d330ad3a07afd052110e67cb8901c820bb0e5181054b5d79a2 +size 5169475119 diff --git a/Compiled/Windows/ApplioV3.6.1.zip b/Compiled/Windows/ApplioV3.6.1.zip new file mode 100644 index 0000000000000000000000000000000000000000..275144d39ceb1a47bb3aa3c5d88192e791988869 --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.1.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c4f3a2eb8b0de1be5f1bf0d34f7437855eaf339c68ee7697d408c8e9cba6fc1 +size 5475095742 diff --git a/Compiled/Windows/ApplioV3.6.2.zip b/Compiled/Windows/ApplioV3.6.2.zip new file mode 100644 index 0000000000000000000000000000000000000000..de0b71bcb8ef06d2ff98d8a2bd55a3a0e7c3d6f4 --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.2.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0763606b8e0b2e2b75077eac5af2772a1682166a812acd9475622fe4c6fb1df9 +size 5440885048 diff --git a/Compiled/Windows/ApplioV3.6.3.zip b/Compiled/Windows/ApplioV3.6.3.zip new file mode 100644 index 0000000000000000000000000000000000000000..5484acd611108ddecd4d6130ae181500c4043aa9 --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.3.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3fa245b6456a8ca7336fd83e64edcb6679969df4c6576ffd899c4ee17699012 +size 5516943808 diff --git a/Compiled/Windows/ApplioV3.6.4.zip b/Compiled/Windows/ApplioV3.6.4.zip new file mode 100644 index 0000000000000000000000000000000000000000..24e63231144480c0d100917e3b87126046c0533b --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.4.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:beedb7dbe54bcc0ccafacbd59fa4ed85af1839c591a84f82b93b90e5109dd9f1 +size 4937966558 diff --git a/Compiled/Windows/ApplioV3.6.5.zip b/Compiled/Windows/ApplioV3.6.5.zip new file mode 100644 index 0000000000000000000000000000000000000000..e72a538f013685d1bb7ffcf6637b0200ae8b8aee --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.5.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d6d777a2668a6b16e83d7648f227961088a54cfcb93025f63099cd1088e1ff2 +size 4929106166 diff --git a/Compiled/Windows/ApplioV3.6.6.zip b/Compiled/Windows/ApplioV3.6.6.zip new file mode 100644 index 0000000000000000000000000000000000000000..b966cad6bbdd616b4ad35604f9425d14107e9497 --- /dev/null +++ b/Compiled/Windows/ApplioV3.6.6.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53526ef4be53e18cd656ff0a51d687416b52129342bc7fdaf481bb71c1fedced +size 6780746968 diff --git a/Compiled/Windows/ApplioV3.7.0.zip b/Compiled/Windows/ApplioV3.7.0.zip new file mode 100644 index 0000000000000000000000000000000000000000..93014b53ea5e28d4cb6732dc7826f92dcb0c2ca2 --- /dev/null +++ b/Compiled/Windows/ApplioV3.7.0.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7eeb07c8fedfde752f83b8a7716cc14645e01d1770f12fd0b392482d3532aeb2 +size 6775972339 diff --git a/Compiled/Windows/ApplioV3.7.1.zip b/Compiled/Windows/ApplioV3.7.1.zip new file mode 100644 index 0000000000000000000000000000000000000000..5ad7209019dd385cf4665ecbe399982847de8120 --- /dev/null +++ b/Compiled/Windows/ApplioV3.7.1.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fcaed5f24814194192d7853b9759cd553d292a4cd4f53f8dbe4e44ac3464de8 +size 6777319254 diff --git a/Enviroment/.gitignore b/Enviroment/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Enviroment/Colab/.gitignore b/Enviroment/Colab/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Enviroment/Colab/Pnpur.tar.gz b/Enviroment/Colab/Pnpur.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..cc52864deae05328c942fc31774d60861e817760 --- /dev/null +++ b/Enviroment/Colab/Pnpur.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edc00278eee827bf24237492b2a91ccdafdb98393a41ec9b4fff9d9a62b5abe6 +size 2529458747 diff --git a/Enviroment/Colab/Pnpur_.tar.gz b/Enviroment/Colab/Pnpur_.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..6293cc7a563b36ee1273455d50e4dde90def65af --- /dev/null +++ b/Enviroment/Colab/Pnpur_.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa42df46fbb0f7e9fad7d493a0ed317754920703be9d148122d766b4aef48ccf +size 3036863135 diff --git a/Enviroment/Colab/tmp_packages.tar.gz b/Enviroment/Colab/tmp_packages.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..33ded9f6aa15a480437ceabf41ca405fd2c2adf7 --- /dev/null +++ b/Enviroment/Colab/tmp_packages.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df6859c5c7f37bb664926db714bf68a9c2b909f03fa84644cddb1fa3b51a8682 +size 30779249 diff --git a/Enviroment/Kaggle/.gitignore b/Enviroment/Kaggle/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/Enviroment/Kaggle/Kaggle.tar.gz b/Enviroment/Kaggle/Kaggle.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..9caaf7602f8cfe27341fe838563a4862c3b0490c --- /dev/null +++ b/Enviroment/Kaggle/Kaggle.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:221e9e89d15775a71f2458b30dee22c409f01e0ca9a9463fea43234c7dabc784 +size 2909878809 diff --git a/Enviroment/Kaggle/KaggleV2.tar.gz b/Enviroment/Kaggle/KaggleV2.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..39477eb9bfc75bbe60ec7615a2eb9251a17d0f57 --- /dev/null +++ b/Enviroment/Kaggle/KaggleV2.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1c19a66ca46bf47f56dd6e46633069b51997d7acd295e36c2e82dd87adc45d8 +size 3249845631 diff --git a/Enviroment/Kaggle/bin_kaggle.tar.gz b/Enviroment/Kaggle/bin_kaggle.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..b7a7dccf80160e239647bfbbef5651367af30633 --- /dev/null +++ b/Enviroment/Kaggle/bin_kaggle.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73e273205da615c7ebda701125fbfe08a701fa26d0658e958001c51c69197644 +size 10203128 diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..39f8a49da238b87ba001316e3f7fa612728d3be2 --- /dev/null +++ b/README.md @@ -0,0 +1,100 @@ +--- +pipeline_tag: audio-to-audio +tags: +- AI +- RVC +- VITS +- VC +- Voice Conversion +- Voice2Voice +license: mit +datasets: +- CSTR-Edinburgh/vctk +base_model: +- lj1995/VoiceConversionWebUI +--- + +

+ Applio +

+ +

A simple, high-quality voice conversion tool, focused on ease of use and performance.

+ +

+ 🌐 Website + • + 📚 Documentation + • + ☎️ Discord +

+ +

+ 🛒 Plugins + • + 📦 Compiled + • + 🎮 Playground + • + 🔎 Google Colab (UI) + • + 🔎 Google Colab (No UI) +

+ +## Introduction + +Applio is a powerful voice conversion tool focused on simplicity, quality, and performance. Whether you're an artist, developer, or researcher, Applio offers a straightforward platform for high-quality voice transformations. Its flexible design allows for customization through plugins and configurations, catering to a wide range of projects. + +## Terms of Use +The use of Applio is entirely at your own discretion and responsibility. By using this tool, you agree to: + +1. Respect all applicable copyrights, intellectual property rights, and privacy rights. Ensure that any audio or material processed through Applio is either owned by you or used with explicit permission from the rightful owner. + +2. Avoid using Applio in ways that may harm, defame, or infringe upon the rights of others. This includes, but is not limited to, the creation or distribution of unauthorized content. + +3. Comply with all relevant laws and regulations governing the use of AI and voice transformation tools in your jurisdiction. + +Applio and its contributors are not liable for any misuse of the tool. The responsibility for adhering to ethical practices and legal compliance lies solely with the user. Applio does not endorse or support any activities that result in harm to individuals, groups, or entities. All official models distributed by Applio have been trained under public use datasets such as VCTK. + +## Getting Started + +### 1. Installation + +Run the installation script based on your operating system: + +- **Windows:** Double-click `run-install.bat`. +- **Linux/macOS:** Execute `run-install.sh`. + +### 2. Running Applio + +Start Applio using: + +- **Windows:** Double-click `run-applio.bat`. +- **Linux/macOS:** Run `run-applio.sh`. + +This launches the Gradio interface in your default browser. + +### 3. Optional: TensorBoard Monitoring + +To monitor training or visualize data: + +- **Windows:** Run `run-tensorboard.bat`. +- **Linux/macOS:** Run `run-tensorboard.sh`. + +For more detailed instructions, visit the [documentation](https://docs.applio.org). + +## Commercial Usage + +For commercial use, follow the [MIT license](./LICENSE) and contact us at support@applio.org to ensure ethical use. The use of Applio-generated audio files must comply with applicable copyrights. Consider supporting Applio’s development [through a donation](https://ko-fi.com/iahispano). + +## References + +Applio is made possible thanks to these projects and their references: + +- [gradio-screen-recorder](https://huggingface.co/spaces/gstaff/gradio-screen-recorder) by gstaff +- [rvc-cli](https://github.com/blaisewf/rvc-cli) by blaisewf + +### Contributors + + + + \ No newline at end of file diff --git a/Resources/contentvec_base.pt b/Resources/contentvec_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f47ab58564f01d5cc8b05c63bdf96d944551ff --- /dev/null +++ b/Resources/contentvec_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96 +size 189507909 diff --git a/Resources/embedders/chinese_hubert_base.pt b/Resources/embedders/chinese_hubert_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee11bf471f035ac8178a874bf19565abf22c92e4 --- /dev/null +++ b/Resources/embedders/chinese_hubert_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cd5db6302ae2e79b5972cd02ae375a42a76170374d6e1952fa78d1fe4e4f756 +size 1136482241 diff --git a/Resources/embedders/chinese_hubert_base/config.json b/Resources/embedders/chinese_hubert_base/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ff578cc6320e2cea57f4fa2f15809e8a6dc3d466 --- /dev/null +++ b/Resources/embedders/chinese_hubert_base/config.json @@ -0,0 +1,71 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModel" + ], + "attention_dropout": 0.1, + "bos_token_id": 1, + "classifier_proj_size": 256, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_norm": "group", + "feat_proj_dropout": 0.0, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "torch_dtype": "float32", + "transformers_version": "4.20.0.dev0", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/chinese_hubert_base/pytorch_model.bin b/Resources/embedders/chinese_hubert_base/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..6451c85db1734fec15769a324311c53c22601597 --- /dev/null +++ b/Resources/embedders/chinese_hubert_base/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fefccd26c2794a583b80f6f7210c721873cb7ebae2c1cde3baf9b27855e24d8 +size 377552987 diff --git a/Resources/embedders/contentvec/config.json b/Resources/embedders/contentvec/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5186a71b15933aca2d9942db95e1aff02642d1f0 --- /dev/null +++ b/Resources/embedders/contentvec/config.json @@ -0,0 +1,71 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModelWithFinalProj" + ], + "attention_dropout": 0.1, + "bos_token_id": 1, + "classifier_proj_size": 256, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_norm": "group", + "feat_proj_dropout": 0.0, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "torch_dtype": "float32", + "transformers_version": "4.27.3", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/contentvec/pytorch_model.bin b/Resources/embedders/contentvec/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..f03cfb9aa43106344342ef928c119d5ad5f9aa5c --- /dev/null +++ b/Resources/embedders/contentvec/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e +size 378342945 diff --git a/Resources/embedders/contentvec_base.pt b/Resources/embedders/contentvec_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f47ab58564f01d5cc8b05c63bdf96d944551ff --- /dev/null +++ b/Resources/embedders/contentvec_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96 +size 189507909 diff --git a/Resources/embedders/hubert_base.pt b/Resources/embedders/hubert_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f47ab58564f01d5cc8b05c63bdf96d944551ff --- /dev/null +++ b/Resources/embedders/hubert_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96 +size 189507909 diff --git a/Resources/embedders/japanese_hubert_base.pt b/Resources/embedders/japanese_hubert_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..e32292b1e3318259ec0d56c6c646f6a82064b2ba --- /dev/null +++ b/Resources/embedders/japanese_hubert_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dade3cf824ae0d214f7de8b73e70bae7c101e81f12d93577c4760bf516db4063 +size 378888853 diff --git a/Resources/embedders/japanese_hubert_base/config.json b/Resources/embedders/japanese_hubert_base/config.json new file mode 100644 index 0000000000000000000000000000000000000000..32a7194b95785e704dcbf916ad807bc6a34027bb --- /dev/null +++ b/Resources/embedders/japanese_hubert_base/config.json @@ -0,0 +1,71 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModel" + ], + "attention_dropout": 0.1, + "bos_token_id": 1, + "classifier_proj_size": 256, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_norm": "group", + "feat_proj_dropout": 0.0, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "torch_dtype": "float32", + "transformers_version": "4.28.1", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/japanese_hubert_base/pytorch_model.bin b/Resources/embedders/japanese_hubert_base/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..74ee7ad0f9d840abe3a37d705f7135012315d57d --- /dev/null +++ b/Resources/embedders/japanese_hubert_base/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c023ccb71e4c2b5a324c94fc5ebe12403d3081c5f370df229892419996fd113 +size 377554841 diff --git a/Resources/embedders/korean_hubert_base.pt b/Resources/embedders/korean_hubert_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..bf13a33f4f8fa20105283d0e64ced545bd8bb023 --- /dev/null +++ b/Resources/embedders/korean_hubert_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b42c8453b96b203198c1c280a8821158ea3fa8dbbc2a6220cad1c1489c3e65e +size 378876997 diff --git a/Resources/embedders/korean_hubert_base/config.json b/Resources/embedders/korean_hubert_base/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5a62087653d090a059e804c216a7d50532258e9f --- /dev/null +++ b/Resources/embedders/korean_hubert_base/config.json @@ -0,0 +1,78 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModel" + ], + "auto_map": { + "FlaxAutoModel": "modeling_flax_hubert.FlaxHubertModel" + }, + "attention_dropout": 0.1, + "bos_token_id": 1, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_dropout": 0.0, + "feat_extract_norm": "group", + "feat_proj_dropout": 0.1, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "final_proj_size": 256, + "gradient_checkpointing": false, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_dropout_prob": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "tokenizer_class": "Wav2Vec2CTCTokenizer", + "torch_dtype": "float32", + "transformers_version": "4.30.0.dev0", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/korean_hubert_base/pytorch_model.bin b/Resources/embedders/korean_hubert_base/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..ba9199bc7c27b23850daf128ca1b6cd1ca133f59 --- /dev/null +++ b/Resources/embedders/korean_hubert_base/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:931f6232879f8eadf7dbd9e00e1fa4cac61ad269af89d509b2ed75009b1a02c5 +size 377554841 diff --git a/Resources/embedders/spin-v2/config.json b/Resources/embedders/spin-v2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5a86e3b44fe9b3b16f265204eeb3cf64afefceaf --- /dev/null +++ b/Resources/embedders/spin-v2/config.json @@ -0,0 +1,72 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModelWithFinalProj" + ], + "attention_dropout": 0.1, + "bos_token_id": 1, + "classifier_proj_size": 256, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_pos_batch_norm": false, + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_norm": "group", + "feat_proj_dropout": 0.0, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "torch_dtype": "float32", + "transformers_version": "4.55.0", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/spin-v2/pytorch_model.bin b/Resources/embedders/spin-v2/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..f1fcd1c9352846f6330d6a215f1bbefa3a8d39cd --- /dev/null +++ b/Resources/embedders/spin-v2/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a9ac0be326057b17607a988be497793817f8274e987cf691a1b61192510f823 +size 378356791 diff --git a/Resources/embedders/spin/config.json b/Resources/embedders/spin/config.json new file mode 100644 index 0000000000000000000000000000000000000000..787124b72efac92eda06941d2b3a36f096820a86 --- /dev/null +++ b/Resources/embedders/spin/config.json @@ -0,0 +1,71 @@ +{ + "activation_dropout": 0.1, + "apply_spec_augment": true, + "architectures": [ + "HubertModelWithFinalProj" + ], + "attention_dropout": 0.1, + "bos_token_id": 1, + "classifier_proj_size": 256, + "conv_bias": false, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "sum", + "ctc_zero_infinity": false, + "do_stable_layer_norm": false, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_norm": "group", + "feat_proj_dropout": 0.0, + "feat_proj_layer_norm": true, + "final_dropout": 0.1, + "hidden_act": "gelu", + "hidden_dropout": 0.1, + "hidden_size": 768, + "initializer_range": 0.02, + "intermediate_size": 3072, + "layer_norm_eps": 1e-05, + "layerdrop": 0.1, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "model_type": "hubert", + "num_attention_heads": 12, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 12, + "pad_token_id": 0, + "torch_dtype": "float32", + "transformers_version": "4.44.2", + "use_weighted_layer_sum": false, + "vocab_size": 32 +} diff --git a/Resources/embedders/spin/pytorch_model.bin b/Resources/embedders/spin/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..58b4f5101f8561f3658a1dd97e8f249dfbfe5e23 --- /dev/null +++ b/Resources/embedders/spin/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:057f12bfda54e2d486d86a52a3beb2a07c96a888bc6ac0c382c12ac18dbd500c +size 378356791 diff --git a/Resources/fcpe.pt b/Resources/fcpe.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ee56f5b0321c8b574d5a4cc69eb0e484522dab9 --- /dev/null +++ b/Resources/fcpe.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8544427eebbf2baef6213cc9a05057e46961617a8e5bd96975a0d42da6a09059 +size 43362881 diff --git a/Resources/ffmpeg.exe b/Resources/ffmpeg.exe new file mode 100644 index 0000000000000000000000000000000000000000..ddf3337846e74afb16a20a0f5405f10072199a06 --- /dev/null +++ b/Resources/ffmpeg.exe @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6a4d917a444790f4c06ada640c1c0c95aecde2f8953ed8d0dfb19352500bfcd +size 52925440 diff --git a/Resources/ffprobe.exe b/Resources/ffprobe.exe new file mode 100644 index 0000000000000000000000000000000000000000..0cf5cc2a4e1e28d32820c0709996b386531fbde0 --- /dev/null +++ b/Resources/ffprobe.exe @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2da5b980a9a14a808f423d181c4ed51c2b8af11b1366699f3f7eab0609926f8f +size 122135040 diff --git a/Resources/formant/stftpitchshift b/Resources/formant/stftpitchshift new file mode 100644 index 0000000000000000000000000000000000000000..4f62e31529b9aafede1b7c08de1f41b980fdf132 --- /dev/null +++ b/Resources/formant/stftpitchshift @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb2f50ea8e5ca1a11a587f11f25ba9182f9b24e2367ac480f430b3f04062782e +size 1822104 diff --git a/Resources/formant/stftpitchshift.exe b/Resources/formant/stftpitchshift.exe new file mode 100644 index 0000000000000000000000000000000000000000..2c7d3d6595d5411fb85813bbeaecd0efc744047a Binary files /dev/null and b/Resources/formant/stftpitchshift.exe differ diff --git a/Resources/formant/stftpitchshift.lib b/Resources/formant/stftpitchshift.lib new file mode 100644 index 0000000000000000000000000000000000000000..9a07fccaf1b2bbc79fb1978b111e4e7ebb3750a1 Binary files /dev/null and b/Resources/formant/stftpitchshift.lib differ diff --git a/Resources/hubert_base.pt b/Resources/hubert_base.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f47ab58564f01d5cc8b05c63bdf96d944551ff --- /dev/null +++ b/Resources/hubert_base.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96 +size 189507909 diff --git a/Resources/predictors/fcpe.pt b/Resources/predictors/fcpe.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ee56f5b0321c8b574d5a4cc69eb0e484522dab9 --- /dev/null +++ b/Resources/predictors/fcpe.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8544427eebbf2baef6213cc9a05057e46961617a8e5bd96975a0d42da6a09059 +size 43362881 diff --git a/Resources/predictors/rmvpe.pt b/Resources/predictors/rmvpe.pt new file mode 100644 index 0000000000000000000000000000000000000000..6362f060846875c3b5d7012adea5f97e47305e7e --- /dev/null +++ b/Resources/predictors/rmvpe.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193 +size 181184272 diff --git a/Resources/predictors/swift.onnx b/Resources/predictors/swift.onnx new file mode 100644 index 0000000000000000000000000000000000000000..84bb1297c4b6a1b389f293816343dbdf9fd8d2dd --- /dev/null +++ b/Resources/predictors/swift.onnx @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6385e8c2ebc3872e82c9ff5946228de44cd3be77a750ea53698b7dbfe94b0a22 +size 135090 diff --git a/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kD.pth b/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kD.pth new file mode 100644 index 0000000000000000000000000000000000000000..bd654fa821b919e59ee189c989253576cac35149 --- /dev/null +++ b/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kD.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff8731f4e11655454e53746fbcd0245a65eedb41e2e34db67f6b06d363a66b78 +size 857141306 diff --git a/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kG.pth b/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kG.pth new file mode 100644 index 0000000000000000000000000000000000000000..183407133cb77614d1f7f7a86cc94ee5873386cc --- /dev/null +++ b/Resources/pretrained_custom/Ov2Super/f0Ov2Super32kG.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:377272acbb7d6ec494c5964a0469e0b57189988cdba734e1bfcbeaaa923b7bc2 +size 443282322 diff --git a/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kD.pth b/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kD.pth new file mode 100644 index 0000000000000000000000000000000000000000..4ccaa88dfeecfcc3032424a4ade160add03528e3 --- /dev/null +++ b/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kD.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e319e21eb26137803c62847857202bc43f833c0696be72bec395a74b3c2178aa +size 857126469 diff --git a/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kG.pth b/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kG.pth new file mode 100644 index 0000000000000000000000000000000000000000..ace9f8065176a23829e521925905b987b50ec016 --- /dev/null +++ b/Resources/pretrained_custom/Ov2Super/f0Ov2Super40kG.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0562bdf6fa73197a503ceddf376a79f33727d64e19b1c6371ebfd6872bceccbf +size 438183069 diff --git a/Resources/pretrained_custom/RIN/f0RIN40kD.pth b/Resources/pretrained_custom/RIN/f0RIN40kD.pth new file mode 100644 index 0000000000000000000000000000000000000000..3f9d8a81f034e73a4a2984f84e9a2e48d159c807 --- /dev/null +++ b/Resources/pretrained_custom/RIN/f0RIN40kD.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0dbe352663c7cd58a131a4516a95066a139474c6304e87dc4d385a8222e6253 +size 857126469 diff --git a/Resources/pretrained_custom/RIN/f0RIN40kG.pth b/Resources/pretrained_custom/RIN/f0RIN40kG.pth new file mode 100644 index 0000000000000000000000000000000000000000..c0a74323c8c4edee25d04881219082eed95c983a --- /dev/null +++ b/Resources/pretrained_custom/RIN/f0RIN40kG.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:814de985f6dc638f0ee6270b0e153cdf63e1c78424d210f12efe03bdb4ea62a7 +size 438183069 diff --git a/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kD.pth b/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kD.pth new file mode 100644 index 0000000000000000000000000000000000000000..a0590b7afd6bbce4891d62b806c967b280cbe7d2 --- /dev/null +++ b/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kD.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dcdcd570ad100d96236f0a77afa997ef4c954adfb7247deb502aa1a6c64fa22 +size 857126469 diff --git a/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kG.pth b/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kG.pth new file mode 100644 index 0000000000000000000000000000000000000000..c0593b19940e41bc781f30cc54d9832aee59ffd4 --- /dev/null +++ b/Resources/pretrained_custom/SnowieRu/f0SnowieRu40kG.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c46c319ac02e46d2997cdf8030a24727f9685b5f7f2b47ae5179a487b454077 +size 438183069 diff --git a/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kD.pth b/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kD.pth new file mode 100644 index 0000000000000000000000000000000000000000..deadf743e00df201a3527e8b7c159e3ba3df841d --- /dev/null +++ b/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kD.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:434c96c89a9b3eebb32ece5c6d7199c4e3ef7df1b91bbbb9d08b3e36c24f6574 +size 857126469 diff --git a/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kG.pth b/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kG.pth new file mode 100644 index 0000000000000000000000000000000000000000..374c190b2de2d8f0d84ef6f090b065e63cce49f9 --- /dev/null +++ b/Resources/pretrained_custom/SnowieRu/f0SnowieRu48kG.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0094b15566a1a9cc69298694fa82fd78c165f3ecab01cf5ee29cd1f66ee91dfb +size 452338845 diff --git a/Resources/pretrained_v1/D32k.pth b/Resources/pretrained_v1/D32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..346293c7023f67178a4c85da94cc3e85bdb320ef --- /dev/null +++ b/Resources/pretrained_v1/D32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ab20645829460fdad0d3c44254f1ab53c32cae50c22a66c926ae5aa30abda6f +size 109978943 diff --git a/Resources/pretrained_v1/D40k.pth b/Resources/pretrained_v1/D40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..11fd6ad63daeb5d786e78d843a6f0800c3d2394c --- /dev/null +++ b/Resources/pretrained_v1/D40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:547f66dbbcd9023b9051ed244d12ab043ba8a4e854b154cc28761ac7c002909b +size 109978943 diff --git a/Resources/pretrained_v1/D48k.pth b/Resources/pretrained_v1/D48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..48beafd328bedd8cf493bbf5f4a9d37d2260024a --- /dev/null +++ b/Resources/pretrained_v1/D48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cc013fa60ed9c3f902f5bd99f48c7e3b9352d763d4d3cd6bc241c37b0bfd9ad +size 109978943 diff --git a/Resources/pretrained_v1/G32k.pth b/Resources/pretrained_v1/G32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..e64d79fd69b7287cdac27cd31dbdfd62c563e094 --- /dev/null +++ b/Resources/pretrained_v1/G32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81817645cde7ed2e2d83f23ef883f33dda564924b497e84d792743912eca4c23 +size 72653893 diff --git a/Resources/pretrained_v1/G40k.pth b/Resources/pretrained_v1/G40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..16796af627338b8c55934df239be3b9d27f9f060 --- /dev/null +++ b/Resources/pretrained_v1/G40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e428573bda1124b0ae0ae843fd8dcded6027d3993444790b3e9b0100938b2113 +size 72763063 diff --git a/Resources/pretrained_v1/G48k.pth b/Resources/pretrained_v1/G48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..509a033306a137648a4f59e6016097b7bd318dd8 --- /dev/null +++ b/Resources/pretrained_v1/G48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3862a67ea6313e8ffefc05cee6bee656ef3e089442e9ecf4a6618d60721f3e95 +size 72850501 diff --git a/Resources/pretrained_v1/f0D32k.pth b/Resources/pretrained_v1/f0D32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..9f36ffb9c4afdf77440f4619688d985daa25b107 --- /dev/null +++ b/Resources/pretrained_v1/f0D32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:294db3087236e2c75260d6179056791c9231245daf5d0485545d9e54c4057c77 +size 109978943 diff --git a/Resources/pretrained_v1/f0D40k.pth b/Resources/pretrained_v1/f0D40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..da7dad91a825dac3a2ed0fb418c99be4a804811e --- /dev/null +++ b/Resources/pretrained_v1/f0D40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d4f5a441594b470d67579958b2fd4c6b992852ded28ff9e72eda67abcebe423 +size 109978943 diff --git a/Resources/pretrained_v1/f0D48k.pth b/Resources/pretrained_v1/f0D48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..789032e0a223731da818f55c6f65df22a354d6e3 --- /dev/null +++ b/Resources/pretrained_v1/f0D48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b84c8bf347ad1e539c842e8f2a4c36ecd9e7fb23c16041189e4877e9b07925c +size 109978943 diff --git a/Resources/pretrained_v1/f0G32k.pth b/Resources/pretrained_v1/f0G32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..b425a989fbdccb9821e512e9c0b4d84b80169769 --- /dev/null +++ b/Resources/pretrained_v1/f0G32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:285f524bf48bb692c76ad7bd0bc654c12bd9e5edeb784dddf7f61a789a608574 +size 72795627 diff --git a/Resources/pretrained_v1/f0G40k.pth b/Resources/pretrained_v1/f0G40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..a4207034410f808123cfef2ea4aeb2d412f15f72 --- /dev/null +++ b/Resources/pretrained_v1/f0G40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9115654aeef1995f7dd3c6fc4140bebbef0ca9760bed798105a2380a34299831 +size 72909665 diff --git a/Resources/pretrained_v1/f0G48k.pth b/Resources/pretrained_v1/f0G48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..8da740c01aeb6ec0054b221fa7da4242139b273b --- /dev/null +++ b/Resources/pretrained_v1/f0G48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78bc9cab27e34bcfc194f93029374d871d8b3e663ddedea32a9709e894cc8fe8 +size 73008619 diff --git a/Resources/pretrained_v2/D32k.pth b/Resources/pretrained_v2/D32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..b55340b0bc2875f148b245e781d534c5dd4f7894 --- /dev/null +++ b/Resources/pretrained_v2/D32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8043378cc6619083d385f5a045de09b83fb3bf8de45c433ca863b71723ac3ca +size 142875703 diff --git a/Resources/pretrained_v2/D40k.pth b/Resources/pretrained_v2/D40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..6d13aea9208310573b59309a9c80310ef71c5547 --- /dev/null +++ b/Resources/pretrained_v2/D40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:471378e894e7191f89a94eda8288c5947b16bbe0b10c3f1f17efdb7a1d998242 +size 142875703 diff --git a/Resources/pretrained_v2/D48k.pth b/Resources/pretrained_v2/D48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..2f3b46950856921425d3dd2a0857ce14fc7dc622 --- /dev/null +++ b/Resources/pretrained_v2/D48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db01094a93c09868a278e03dafe8bb781bfcc1a5ba8df168c948bf9168c84d82 +size 142875703 diff --git a/Resources/pretrained_v2/G32k.pth b/Resources/pretrained_v2/G32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..d34c78904a383344a31fa9814782e55d82d852ae --- /dev/null +++ b/Resources/pretrained_v2/G32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:869b26a47f75168d6126f64ac39e6de5247017a8658cfd68aca600f7323efb9f +size 73811639 diff --git a/Resources/pretrained_v2/G40k.pth b/Resources/pretrained_v2/G40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..ee39bf64a1fc1d0d8154e242a3b60ef3e2abf0ca --- /dev/null +++ b/Resources/pretrained_v2/G40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3843da7fde33db1dab176146c70d6c2df06eafe9457f4e3aa10024e9c6a4b69 +size 72959671 diff --git a/Resources/pretrained_v2/G48k.pth b/Resources/pretrained_v2/G48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..71fb5f1037137dae5ae4a5786c452197f3eabab1 --- /dev/null +++ b/Resources/pretrained_v2/G48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e2b1581a436d07a76b10b9d38765f64aa02836dc65c7dee1ce4140c11ea158b +size 75318967 diff --git a/Resources/pretrained_v2/f0D32k.pth b/Resources/pretrained_v2/f0D32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..7449cc83c3f97fd87c62596f005d262120b64772 --- /dev/null +++ b/Resources/pretrained_v2/f0D32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd7134e7793674c85474d5145d2d982e3c5d8124fc7bb6c20f710ed65808fa8a +size 142875703 diff --git a/Resources/pretrained_v2/f0D40k.pth b/Resources/pretrained_v2/f0D40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..68e29fb9debf8994f68cb4f31cc4d81d360fb73b --- /dev/null +++ b/Resources/pretrained_v2/f0D40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b6ab091e70801b28e3f41f335f2fc5f3f35c75b39ae2628d419644ec2b0fa09 +size 142875703 diff --git a/Resources/pretrained_v2/f0D48k.pth b/Resources/pretrained_v2/f0D48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..a9f2b74cdccf9bfc2ebb0be8b481c76552a28661 --- /dev/null +++ b/Resources/pretrained_v2/f0D48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2269b73c7a4cf34da09aea99274dabf99b2ddb8a42cbfb065fb3c0aa9a2fc748 +size 142875703 diff --git a/Resources/pretrained_v2/f0G32k.pth b/Resources/pretrained_v2/f0G32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..6bff869bb04fae663dddd1fbe5c04b497bfc4209 --- /dev/null +++ b/Resources/pretrained_v2/f0G32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2332611297b8d88c7436de8f17ef5f07a2119353e962cd93cda5806d59a1133d +size 73950049 diff --git a/Resources/pretrained_v2/f0G40k.pth b/Resources/pretrained_v2/f0G40k.pth new file mode 100644 index 0000000000000000000000000000000000000000..3333a970abc908b4e8afd45c1cc2120ce1b4b0b4 --- /dev/null +++ b/Resources/pretrained_v2/f0G40k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b2c44035e782c4b14ddc0bede9e2f4a724d025cd073f736d4f43708453adfcb +size 73106273 diff --git a/Resources/pretrained_v2/f0G48k.pth b/Resources/pretrained_v2/f0G48k.pth new file mode 100644 index 0000000000000000000000000000000000000000..34d1ec0e2594379989cad7b72991152a9f97d99e --- /dev/null +++ b/Resources/pretrained_v2/f0G48k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5d51f589cc3632d4eae36a315b4179397695042edc01d15312e1bddc2b764a4 +size 75465569 diff --git a/Resources/refinegan/f0D24k.pth b/Resources/refinegan/f0D24k.pth new file mode 100644 index 0000000000000000000000000000000000000000..9074f97f85de0cbcd892666c3548612e42ac3fdc --- /dev/null +++ b/Resources/refinegan/f0D24k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2207181b3fe6d91bb54355ab9822905e64f3e40f3793045f770bfb407a1d5f3 +size 188158474 diff --git a/Resources/refinegan/f0D32k.pth b/Resources/refinegan/f0D32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..c432c676baade642a1b43ec2d462e88dc6f3cac1 --- /dev/null +++ b/Resources/refinegan/f0D32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3156102c0578a5c4a426a00e454abff274f287f4146c9db757546d36544296f7 +size 188166756 diff --git a/Resources/refinegan/f0G24k.pth b/Resources/refinegan/f0G24k.pth new file mode 100644 index 0000000000000000000000000000000000000000..fb2f4224b9f67aca30a6243962f339f15340f5ad --- /dev/null +++ b/Resources/refinegan/f0G24k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:182901b02c47fd065a885d7f0b4b1256d8667373d687434e840a557fb4d3d5e3 +size 140589653 diff --git a/Resources/refinegan/f0G32k.pth b/Resources/refinegan/f0G32k.pth new file mode 100644 index 0000000000000000000000000000000000000000..af818adf2a6a0ffe9a7a5872a7ca29ff402d7f7d --- /dev/null +++ b/Resources/refinegan/f0G32k.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:daad0a22ade3a1f206a65432ffe1705d090631b856d46781a8703e4ce0203484 +size 140604497 diff --git a/Resources/rmvpe.onnx b/Resources/rmvpe.onnx new file mode 100644 index 0000000000000000000000000000000000000000..e080c08fc35bbe5c872546ef2b2fb05fa7192f99 --- /dev/null +++ b/Resources/rmvpe.onnx @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5370e71ac80af8b4b7c793d27efd51fd8bf962de3a7ede0766dac0befa3660fd +size 361688443 diff --git a/Resources/rmvpe.pt b/Resources/rmvpe.pt new file mode 100644 index 0000000000000000000000000000000000000000..6362f060846875c3b5d7012adea5f97e47305e7e --- /dev/null +++ b/Resources/rmvpe.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193 +size 181184272 diff --git a/Resources/tts/catalog.json b/Resources/tts/catalog.json new file mode 100644 index 0000000000000000000000000000000000000000..694788a2bb3ad931ed07e52ae89de4e49f0dc44c --- /dev/null +++ b/Resources/tts/catalog.json @@ -0,0 +1,871 @@ +{ + "schemaVersion": 1, + "models": [ + { + "id": "applio-tts-en-us", + "language": "en-us", + "sha256": "45e5bc97122e88b28c54ba6396f0058f9bbc5e46c1adadfd6274b48744ee713b", + "bytes": 23650194, + "format": "applio-text-rvc-frontend-v1", + "config": { + "feature_dim": 768, + "hidden": 256, + "layers": 4, + "reduction": 1, + "architecture": "duration", + "phonetic": true, + "phoneme_vocabulary": { + "": 1, + "": 0, + "": 2, + "": 3, + "n": 4, + "s": 5, + "t": 6, + "ə": 7, + "l": 8, + "a": 9, + "i": 10, + "k": 11, + "d": 12, + "m": 13, + "ɛ": 14, + "ɾ": 15, + "e": 16, + "ɪ": 17, + "p": 18, + "o": 19, + "ɐ": 20, + "z": 21, + "ð": 22, + "f": 23, + "j": 24, + "v": 25, + "b": 26, + "ɹ": 27, + "ʁ": 28, + "ʊ": 29, + "iː": 30, + "r": 31, + "w": 32, + "ʌ": 33, + "u": 34, + "ɡ": 35, + "æ": 36, + "aɪ": 37, + "ʃ": 38, + "h": 39, + "ɔ": 40, + "ɑː": 41, + "ŋ": 42, + "ɚ": 43, + "eɪ": 44, + "β": 45, + "uː": 46, + "y": 47, + "ɑ̃": 48, + "oʊ": 49, + "ᵻ": 50, + "eː": 51, + "θ": 52, + "aʊ": 53, + "ts": 54, + "oː": 55, + "ɔ̃": 56, + "ɣ": 57, + "ɜ": 58, + "ɑ": 59, + "dʒ": 60, + "əl": 61, + "x": 62, + "ɜː": 63, + "ç": 64, + "ʒ": 65, + "tʃ": 66, + "ɔː": 67, + "ɑːɹ": 68, + "ɛ̃": 69, + "ʎ": 70, + "ɔːɹ": 71, + "ʋ": 72, + "aː": 73, + "ɕ": 74, + "œ": 75, + "ø": 76, + "oːɹ": 77, + "ɲ": 78, + "yː": 79, + "ʔ": 80, + "iə": 81, + "i5": 82, + "s.": 83, + "tɕ": 84, + "??": 85, + "nʲ": 86, + "ɛː": 87, + "œ̃": 88, + "ɭ": 89, + "ɔø": 90, + "ʑ": 91, + "tʲ": 92, + "ɨ": 93, + "ɛɹ": 94, + "ts.": 95, + "rʲ": 96, + "ɪɹ": 97, + "ɭʲ": 98, + "i.5": 99, + "ɔɪ": 100, + "q": 101, + "sʲ": 102, + "u5": 103, + "ʊɹ": 104, + "iɜ": 105, + "a5": 106, + "iɛ5": 107, + "øː": 108, + "ʕ": 109, + "ja": 110, + "əɜ": 111, + "th": 112, + "ɑ5": 113, + "oɪ": 114, + "dʲ": 115, + "ə5": 116, + "tɕh": 117, + "ts.h": 118, + "mʲ": 119, + "ɯ": 120, + "dʑ": 121, + "vʲ": 122, + "e̞": 123, + "tʃʲ": 124, + "ei5": 125, + "o5": 126, + "onɡ5": 127, + "ɑu5": 128, + "iɑ5": 129, + "ai5": 130, + "aɪɚ": 131, + "kh": 132, + "ə1": 133, + "ʐ": 134, + "i2": 135, + "ʉ": 136, + "ħ": 137, + "t[": 138, + "aɪə": 139, + "ʲ": 140, + "ju": 141, + "ə2": 142, + "u2": 143, + "oɜ": 144, + "pː": 145, + "iɛɜ": 146, + "ou5": 147, + "y5": 148, + "uɜ": 149, + "tː": 150, + "uo5": 151, + "d[": 152, + "uoɜ": 153, + "tsh": 154, + "ɑɜ": 155, + "ɵ": 156, + "i̪5": 157, + "uei5": 158, + "ɟ": 159, + "aɜ": 160, + "ɑɨ": 161, + "i.ɜ": 162, + "eʊ": 163, + "o2": 164, + "ɐ̃": 165, + "ä": 166, + "pʲ": 167, + "kʲ": 168, + "n̩": 169, + "ɒ": 170, + "ph": 171, + "ɑu2": 172, + "uɨ": 173, + "əɪ": 174, + "ɫ": 175, + "ɬ": 176, + "yɜ": 177, + "bʲ": 178, + "ɑ2": 179, + "s̪": 180, + "aiɜ": 181, + "χ": 182, + "ɐ̃ʊ̃": 183, + "1": 184, + "ə4": 185, + "yæɜ": 186, + "a2": 187, + "ɨː": 188, + "t̪": 189, + "iouɜ": 190, + "ũ": 191, + "onɡɜ": 192, + "aɨ": 193, + "iɛ2": 194, + "ɔɨ": 195, + "ɑuɜ": 196, + "o̞": 197, + "ei2": 198, + "iou2": 199, + "c": 200, + "kː": 201, + "y2": 202, + "ɖ": 203, + "oe": 204, + "dˤ": 205, + "yɛɜ": 206, + "əʊ": 207, + "S": 208, + "ɡʲ": 209, + "onɡ2": 210, + "u\"": 211, + "eiɜ": 212, + "ʈ": 213, + "ɯᵝ": 214, + "iou5": 215, + "dZ": 216, + "r̝̊": 217, + "i.2": 218, + "tS": 219, + "s^": 220, + "ʝ": 221, + "yə5": 222, + "iɑɜ": 223, + "uə5": 224, + "pf": 225, + "ɨu": 226, + "iɑ2": 227, + "ou2": 228, + "ər2": 229, + "fʲ": 230, + "ai2": 231, + "r̝": 232, + "uəɜ": 233, + "ɳ": 234, + "əɨ": 235, + "ua5": 236, + "uɪ": 237, + "ɽ": 238, + "bː": 239, + "yu5": 240, + "uo2": 241, + "yɛ5": 242, + "l̩": 243, + "ɻ": 244, + "ərɜ": 245, + "ʂ": 246, + "i̪2": 247, + "ouɜ": 248, + "uaɜ": 249, + "a.": 250, + "a.ː": 251, + "yæ5": 252, + "dː": 253, + "r̩": 254, + "ee": 255, + "ɪu": 256, + "ər5": 257, + "i̪ɜ": 258, + "æi": 259, + "u:": 260, + "i.ː": 261, + "t^": 262, + "o1": 263, + "ɪ^": 264, + "ai": 265, + "ueiɜ": 266, + "æː": 267, + "ɛɪ": 268, + "eə": 269, + "i.": 270, + "ɴ": 271, + "ie": 272, + "ua2": 273, + "ɑ1": 274, + "o4": 275, + "tʃː": 276, + "o:": 277, + "ɑ:": 278, + "u1": 279, + "N": 280, + "i̪1": 281, + "au": 282, + "yæ2": 283, + "u.": 284, + "qː": 285, + "yəɜ": 286, + "y:": 287, + "kʰ": 288, + "tʃʰ": 289, + "iʊ": 290, + "sx": 291, + "õ": 292, + "uo": 293, + "tʰ": 294, + "uai5": 295, + "bʰ": 296, + "u.ː": 297, + "uə2": 298, + "ʊə": 299, + "d^": 300, + "s̪ː": 301, + "yiɜ": 302, + "dʰ": 303, + "r.": 304, + "oe:": 305, + "i1": 306, + "ɟː": 307, + "yu2": 308, + "nʲʲ": 309, + "i̪4": 310, + "uei2": 311, + "tsʲ": 312, + "ɸ": 313, + "ĩ": 314, + "ɑ4": 315, + "t̪ː": 316, + "eɑ": 317, + "u4": 318, + "e:": 319, + "tsː": 320, + "ʈʰ": 321, + "ɡʰ": 322, + "ɯɯ": 323, + "dʒʲ": 324, + "ʂʲ": 325, + "X": 326, + "ɵː": 327, + "uaiɜ": 328, + "tɕʲ": 329, + "ã": 330, + "t^ː": 331, + "ẽː": 332, + "yɛ2": 333, + "cː": 334, + "i.1": 335, + "ɛʊ": 336, + "dˤdˤ": 337, + "dʒː": 338, + "i4": 339, + "ɡː": 340, + "yi": 341, + "ɕʲ": 342, + "ɟʰ": 343, + "pʰ": 344, + "dʑʲ": 345, + "yuɜ": 346, + "ua1": 347, + "ua4": 348, + "æiː": 349, + "ɐɐ": 350, + "ui": 351, + "iou1": 352, + "ʊː": 353, + "a1": 354, + "iou4": 355, + "cʰ": 356, + "iɛ1": 357, + "yə2": 358, + "ɖʰ": 359, + "ẽ": 360, + "ʒʲ": 361, + "ää": 362, + "ər4": 363, + "iːː": 364, + "ɪː": 365, + "iɑ1": 366, + "ər1": 367, + "œː": 368, + "øi": 369, + "ɪuː": 370, + "cʰcʰ": 371, + "əː1": 372, + "iː1": 373, + "ũ": 374, + "kʰː": 375, + "o̞o̞": 376, + "xʲ": 377, + "ou1": 378, + "iɛ4": 379, + "e̞e̞": 380, + "y1": 381, + "dzː": 382, + "dʲʲ": 383, + "dʰː": 384, + "ɯᵝɯᵝ": 385, + "lː": 386, + "uo1": 387, + "i.4": 388, + "i:": 389, + "yɛ5ʲ": 390, + "a4": 391, + "": 509, + "": 508, + "": 507, + "": 506, + "": 505 + }, + "language": "en-us" + }, + "feature_space": { + "version": "v2", + "embedder_sha256": "d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e", + "embedder_config_sha256": "2ddde063b795d38d9051a7215a092fecf4cfe148b54251e38de51d88d356898b", + "frame_hz": 100, + "feature_dim": 768, + "pitch": "rmvpe", + "rmvpe_sha256": "6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193" + }, + "source_attribution": "LJ Speech 1.1 (Keith Ito and Linda Johnson); public-domain recordings and annotations. https://keithito.com/LJ-Speech-Dataset/", + "initialized_from_sha256": "ede4a46be5e0eddd979848313ef3af65d8836f20f59cc52c29a1e30e24489b81", + "training_manifest_sha256": "09ee901451c3ce0fdf0186a4e5b18c693f8dfc5c95b1a0945944eb041d8025c6", + "license": "CC-BY-4.0", + "license_url": "https://creativecommons.org/licenses/by/4.0/", + "quality_status": "Development checkpoint; naturalness and pitch stability still require listening acceptance", + "url": "https://huggingface.co/IAHispano/Applio/resolve/main/Resources/tts/en-us/frontend.pt", + "package_sha256": "4274ec7fdcf94f03afc76f4024dc3197e1befff469d92004e4a7f7d6c6b2632d", + "file": "en-us/frontend.pt" + }, + { + "id": "applio-tts-es", + "language": "es", + "sha256": "4f8eda19e75634fe438ba232b3a6fa90f11c19ce078748da4b61a5570f4a6c99", + "bytes": 23197714, + "format": "applio-text-rvc-frontend-v1", + "config": { + "feature_dim": 768, + "hidden": 256, + "layers": 4, + "reduction": 1, + "architecture": "duration", + "phonetic": true, + "phoneme_vocabulary": { + "": 1, + "": 0, + "": 2, + "": 3, + "n": 4, + "s": 5, + "t": 6, + "ə": 7, + "l": 8, + "a": 9, + "i": 10, + "k": 11, + "d": 12, + "m": 13, + "ɛ": 14, + "ɾ": 15, + "e": 16, + "ɪ": 17, + "p": 18, + "o": 19, + "ɐ": 20, + "z": 21, + "ð": 22, + "f": 23, + "j": 24, + "v": 25, + "b": 26, + "ɹ": 27, + "ʁ": 28, + "ʊ": 29, + "iː": 30, + "r": 31, + "w": 32, + "ʌ": 33, + "u": 34, + "ɡ": 35, + "æ": 36, + "aɪ": 37, + "ʃ": 38, + "h": 39, + "ɔ": 40, + "ɑː": 41, + "ŋ": 42, + "ɚ": 43, + "eɪ": 44, + "β": 45, + "uː": 46, + "y": 47, + "ɑ̃": 48, + "oʊ": 49, + "ᵻ": 50, + "eː": 51, + "θ": 52, + "aʊ": 53, + "ts": 54, + "oː": 55, + "ɔ̃": 56, + "ɣ": 57, + "ɜ": 58, + "ɑ": 59, + "dʒ": 60, + "əl": 61, + "x": 62, + "ɜː": 63, + "ç": 64, + "ʒ": 65, + "tʃ": 66, + "ɔː": 67, + "ɑːɹ": 68, + "ɛ̃": 69, + "ʎ": 70, + "ɔːɹ": 71, + "ʋ": 72, + "aː": 73, + "ɕ": 74, + "œ": 75, + "ø": 76, + "oːɹ": 77, + "ɲ": 78, + "yː": 79, + "ʔ": 80, + "iə": 81, + "i5": 82, + "s.": 83, + "tɕ": 84, + "??": 85, + "nʲ": 86, + "ɛː": 87, + "œ̃": 88, + "ɭ": 89, + "ɔø": 90, + "ʑ": 91, + "tʲ": 92, + "ɨ": 93, + "ɛɹ": 94, + "ts.": 95, + "rʲ": 96, + "ɪɹ": 97, + "ɭʲ": 98, + "i.5": 99, + "ɔɪ": 100, + "q": 101, + "sʲ": 102, + "u5": 103, + "ʊɹ": 104, + "iɜ": 105, + "a5": 106, + "iɛ5": 107, + "øː": 108, + "ʕ": 109, + "ja": 110, + "əɜ": 111, + "th": 112, + "ɑ5": 113, + "oɪ": 114, + "dʲ": 115, + "ə5": 116, + "tɕh": 117, + "ts.h": 118, + "mʲ": 119, + "ɯ": 120, + "dʑ": 121, + "vʲ": 122, + "e̞": 123, + "tʃʲ": 124, + "ei5": 125, + "o5": 126, + "onɡ5": 127, + "ɑu5": 128, + "iɑ5": 129, + "ai5": 130, + "aɪɚ": 131, + "kh": 132, + "ə1": 133, + "ʐ": 134, + "i2": 135, + "ʉ": 136, + "ħ": 137, + "t[": 138, + "aɪə": 139, + "ʲ": 140, + "ju": 141, + "ə2": 142, + "u2": 143, + "oɜ": 144, + "pː": 145, + "iɛɜ": 146, + "ou5": 147, + "y5": 148, + "uɜ": 149, + "tː": 150, + "uo5": 151, + "d[": 152, + "uoɜ": 153, + "tsh": 154, + "ɑɜ": 155, + "ɵ": 156, + "i̪5": 157, + "uei5": 158, + "ɟ": 159, + "aɜ": 160, + "ɑɨ": 161, + "i.ɜ": 162, + "eʊ": 163, + "o2": 164, + "ɐ̃": 165, + "ä": 166, + "pʲ": 167, + "kʲ": 168, + "n̩": 169, + "ɒ": 170, + "ph": 171, + "ɑu2": 172, + "uɨ": 173, + "əɪ": 174, + "ɫ": 175, + "ɬ": 176, + "yɜ": 177, + "bʲ": 178, + "ɑ2": 179, + "s̪": 180, + "aiɜ": 181, + "χ": 182, + "ɐ̃ʊ̃": 183, + "1": 184, + "ə4": 185, + "yæɜ": 186, + "a2": 187, + "ɨː": 188, + "t̪": 189, + "iouɜ": 190, + "ũ": 191, + "onɡɜ": 192, + "aɨ": 193, + "iɛ2": 194, + "ɔɨ": 195, + "ɑuɜ": 196, + "o̞": 197, + "ei2": 198, + "iou2": 199, + "c": 200, + "kː": 201, + "y2": 202, + "ɖ": 203, + "oe": 204, + "dˤ": 205, + "yɛɜ": 206, + "əʊ": 207, + "S": 208, + "ɡʲ": 209, + "onɡ2": 210, + "u\"": 211, + "eiɜ": 212, + "ʈ": 213, + "ɯᵝ": 214, + "iou5": 215, + "dZ": 216, + "r̝̊": 217, + "i.2": 218, + "tS": 219, + "s^": 220, + "ʝ": 221, + "yə5": 222, + "iɑɜ": 223, + "uə5": 224, + "pf": 225, + "ɨu": 226, + "iɑ2": 227, + "ou2": 228, + "ər2": 229, + "fʲ": 230, + "ai2": 231, + "r̝": 232, + "uəɜ": 233, + "ɳ": 234, + "əɨ": 235, + "ua5": 236, + "uɪ": 237, + "ɽ": 238, + "bː": 239, + "yu5": 240, + "uo2": 241, + "yɛ5": 242, + "l̩": 243, + "ɻ": 244, + "ərɜ": 245, + "ʂ": 246, + "i̪2": 247, + "ouɜ": 248, + "uaɜ": 249, + "a.": 250, + "a.ː": 251, + "yæ5": 252, + "dː": 253, + "r̩": 254, + "ee": 255, + "ɪu": 256, + "ər5": 257, + "i̪ɜ": 258, + "æi": 259, + "u:": 260, + "i.ː": 261, + "t^": 262, + "o1": 263, + "ɪ^": 264, + "ai": 265, + "ueiɜ": 266, + "æː": 267, + "ɛɪ": 268, + "eə": 269, + "i.": 270, + "ɴ": 271, + "ie": 272, + "ua2": 273, + "ɑ1": 274, + "o4": 275, + "tʃː": 276, + "o:": 277, + "ɑ:": 278, + "u1": 279, + "N": 280, + "i̪1": 281, + "au": 282, + "yæ2": 283, + "u.": 284, + "qː": 285, + "yəɜ": 286, + "y:": 287, + "kʰ": 288, + "tʃʰ": 289, + "iʊ": 290, + "sx": 291, + "õ": 292, + "uo": 293, + "tʰ": 294, + "uai5": 295, + "bʰ": 296, + "u.ː": 297, + "uə2": 298, + "ʊə": 299, + "d^": 300, + "s̪ː": 301, + "yiɜ": 302, + "dʰ": 303, + "r.": 304, + "oe:": 305, + "i1": 306, + "ɟː": 307, + "yu2": 308, + "nʲʲ": 309, + "i̪4": 310, + "uei2": 311, + "tsʲ": 312, + "ɸ": 313, + "ĩ": 314, + "ɑ4": 315, + "t̪ː": 316, + "eɑ": 317, + "u4": 318, + "e:": 319, + "tsː": 320, + "ʈʰ": 321, + "ɡʰ": 322, + "ɯɯ": 323, + "dʒʲ": 324, + "ʂʲ": 325, + "X": 326, + "ɵː": 327, + "uaiɜ": 328, + "tɕʲ": 329, + "ã": 330, + "t^ː": 331, + "ẽː": 332, + "yɛ2": 333, + "cː": 334, + "i.1": 335, + "ɛʊ": 336, + "dˤdˤ": 337, + "dʒː": 338, + "i4": 339, + "ɡː": 340, + "yi": 341, + "ɕʲ": 342, + "ɟʰ": 343, + "pʰ": 344, + "dʑʲ": 345, + "yuɜ": 346, + "ua1": 347, + "ua4": 348, + "æiː": 349, + "ɐɐ": 350, + "ui": 351, + "iou1": 352, + "ʊː": 353, + "a1": 354, + "iou4": 355, + "cʰ": 356, + "iɛ1": 357, + "yə2": 358, + "ɖʰ": 359, + "ẽ": 360, + "ʒʲ": 361, + "ää": 362, + "ər4": 363, + "iːː": 364, + "ɪː": 365, + "iɑ1": 366, + "ər1": 367, + "œː": 368, + "øi": 369, + "ɪuː": 370, + "cʰcʰ": 371, + "əː1": 372, + "iː1": 373, + "ũ": 374, + "kʰː": 375, + "o̞o̞": 376, + "xʲ": 377, + "ou1": 378, + "iɛ4": 379, + "e̞e̞": 380, + "y1": 381, + "dzː": 382, + "dʲʲ": 383, + "dʰː": 384, + "ɯᵝɯᵝ": 385, + "lː": 386, + "uo1": 387, + "i.4": 388, + "i:": 389, + "yɛ5ʲ": 390, + "a4": 391, + "": 509, + "": 508, + "": 507, + "": 506, + "": 505 + }, + "language": "es" + }, + "feature_space": { + "version": "v2", + "embedder_sha256": "d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e", + "embedder_config_sha256": "2ddde063b795d38d9051a7215a092fecf4cfe148b54251e38de51d88d356898b", + "frame_hz": 100, + "feature_dim": 768, + "pitch": "rmvpe", + "rmvpe_sha256": "6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193" + }, + "source_attribution": "Multilingual LibriSpeech (Vineel Pratap et al., Meta/Facebook), CC BY 4.0, publisher facebook/multilingual_librispeech revision 2e83e61823b4c47dcbcb1980bb88601274127609, speakers 10246 and 8881. https://huggingface.co/datasets/facebook/multilingual_librispeech . Initialization also descends from the LJ Speech public-domain base.", + "initialized_from_sha256": "bd6b4136d04c0b80cf8d5b790d0c54d6e8a374758c2abd24067070a59647b10a", + "training_manifest_sha256": "4a6567d78f87e36f1c79f8c6296c436bd6238a2c50df7e13ffe10e6f7e05c9d6", + "license": "CC-BY-4.0", + "license_url": "https://creativecommons.org/licenses/by/4.0/", + "quality_status": "Development checkpoint; naturalness and pitch stability still require listening acceptance", + "url": "https://huggingface.co/IAHispano/Applio/resolve/main/Resources/tts/es/frontend.pt", + "package_sha256": "8111c99d84cd840fdea4a91fbe2f6c8ab54566df216d1060f0c15c275c3e94a4", + "file": "es/frontend.pt" + } + ] +} diff --git a/Resources/tts/en-us/README.md b/Resources/tts/en-us/README.md new file mode 100644 index 0000000000000000000000000000000000000000..bcf021789056df8cec538a0aa9464e087e5d1160 --- /dev/null +++ b/Resources/tts/en-us/README.md @@ -0,0 +1,27 @@ +--- +language: +- en-us +license: cc-by-4.0 +tags: +- applio +- text-to-speech +--- +# Applio TTS companion (en-us) + +A duration-based text-to-ContentVec companion. Requires the Applio text-RVC runtime and a compatible RVC v2 voice; this is not an RVC generator/discriminator pretrain or a standalone audio decoder. + +Import `frontend.pt` using Applio's TTS model upload control, then select the same speech language. RVC voices themselves are language agnostic. This checkpoint supplies pronunciation and timing only for its stated language; additional languages are not validated by its existence. + +SHA-256: `45e5bc97122e88b28c54ba6396f0058f9bbc5e46c1adadfd6274b48744ee713b`. Checkpoint file bytes: 23650194. No source recordings, transcripts, recognizers or RVC voice weights are included. + +## Training and attribution + +LJ Speech 1.1 (Keith Ito and Linda Johnson); public-domain recordings and annotations. https://keithito.com/LJ-Speech-Dataset/ + +Architecture, feature-space version and initialization hash are in `model.json`. The checkpoint itself retains training/validation identities and initialization provenance. Applio engine source: https://github.com/IAHispano/Applio . Checkpoint redistribution is offered under CC BY 4.0 with the attribution above: https://creativecommons.org/licenses/by/4.0/ . + +## Limitations and evaluation + +This is a development checkpoint. Intelligibility has been measured on held-out recordings through an unchanged RVC decoder, but natural rhythm and stable pitch have not received human acceptance. Final evaluation sets have already been used and must not be reused for tuning. Automatically generated transcripts can contain errors. These shared bases were trained using paired source transcripts, not automatically transcribed data. + +Current measured comparisons and playable samples: https://github.com/IAHispano/Applio-App/pull/22 . Figures for voice-adapted companions must not be attributed to this shared base. diff --git a/Resources/tts/en-us/frontend.pt b/Resources/tts/en-us/frontend.pt new file mode 100644 index 0000000000000000000000000000000000000000..1e65f62fe99221823b5cc0d959abc82badf495da --- /dev/null +++ b/Resources/tts/en-us/frontend.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45e5bc97122e88b28c54ba6396f0058f9bbc5e46c1adadfd6274b48744ee713b +size 23650194 diff --git a/Resources/tts/en-us/model.json b/Resources/tts/en-us/model.json new file mode 100644 index 0000000000000000000000000000000000000000..44a2f700578976f48dc23ffbfa863305508a000a --- /dev/null +++ b/Resources/tts/en-us/model.json @@ -0,0 +1,431 @@ +{ + "id": "applio-tts-en-us", + "language": "en-us", + "sha256": "45e5bc97122e88b28c54ba6396f0058f9bbc5e46c1adadfd6274b48744ee713b", + "bytes": 23650194, + "format": "applio-text-rvc-frontend-v1", + "config": { + "feature_dim": 768, + "hidden": 256, + "layers": 4, + "reduction": 1, + "architecture": "duration", + "phonetic": true, + "phoneme_vocabulary": { + "": 1, + "": 0, + "": 2, + "": 3, + "n": 4, + "s": 5, + "t": 6, + "ə": 7, + "l": 8, + "a": 9, + "i": 10, + "k": 11, + "d": 12, + "m": 13, + "ɛ": 14, + "ɾ": 15, + "e": 16, + "ɪ": 17, + "p": 18, + "o": 19, + "ɐ": 20, + "z": 21, + "ð": 22, + "f": 23, + "j": 24, + "v": 25, + "b": 26, + "ɹ": 27, + "ʁ": 28, + "ʊ": 29, + "iː": 30, + "r": 31, + "w": 32, + "ʌ": 33, + "u": 34, + "ɡ": 35, + "æ": 36, + "aɪ": 37, + "ʃ": 38, + "h": 39, + "ɔ": 40, + "ɑː": 41, + "ŋ": 42, + "ɚ": 43, + "eɪ": 44, + "β": 45, + "uː": 46, + "y": 47, + "ɑ̃": 48, + "oʊ": 49, + "ᵻ": 50, + "eː": 51, + "θ": 52, + "aʊ": 53, + "ts": 54, + "oː": 55, + "ɔ̃": 56, + "ɣ": 57, + "ɜ": 58, + "ɑ": 59, + "dʒ": 60, + "əl": 61, + "x": 62, + "ɜː": 63, + "ç": 64, + "ʒ": 65, + "tʃ": 66, + "ɔː": 67, + "ɑːɹ": 68, + "ɛ̃": 69, + "ʎ": 70, + "ɔːɹ": 71, + "ʋ": 72, + "aː": 73, + "ɕ": 74, + "œ": 75, + "ø": 76, + "oːɹ": 77, + "ɲ": 78, + "yː": 79, + "ʔ": 80, + "iə": 81, + "i5": 82, + "s.": 83, + "tɕ": 84, + "??": 85, + "nʲ": 86, + "ɛː": 87, + "œ̃": 88, + "ɭ": 89, + "ɔø": 90, + "ʑ": 91, + "tʲ": 92, + "ɨ": 93, + "ɛɹ": 94, + "ts.": 95, + "rʲ": 96, + "ɪɹ": 97, + "ɭʲ": 98, + "i.5": 99, + "ɔɪ": 100, + "q": 101, + "sʲ": 102, + "u5": 103, + "ʊɹ": 104, + "iɜ": 105, + "a5": 106, + "iɛ5": 107, + "øː": 108, + "ʕ": 109, + "ja": 110, + "əɜ": 111, + "th": 112, + "ɑ5": 113, + "oɪ": 114, + "dʲ": 115, + "ə5": 116, + "tɕh": 117, + "ts.h": 118, + "mʲ": 119, + "ɯ": 120, + "dʑ": 121, + "vʲ": 122, + "e̞": 123, + "tʃʲ": 124, + "ei5": 125, + "o5": 126, + "onɡ5": 127, + "ɑu5": 128, + "iɑ5": 129, + "ai5": 130, + "aɪɚ": 131, + "kh": 132, + "ə1": 133, + "ʐ": 134, + "i2": 135, + "ʉ": 136, + "ħ": 137, + "t[": 138, + "aɪə": 139, + "ʲ": 140, + "ju": 141, + "ə2": 142, + "u2": 143, + "oɜ": 144, + "pː": 145, + "iɛɜ": 146, + "ou5": 147, + "y5": 148, + "uɜ": 149, + "tː": 150, + "uo5": 151, + "d[": 152, + "uoɜ": 153, + "tsh": 154, + "ɑɜ": 155, + "ɵ": 156, + "i̪5": 157, + "uei5": 158, + "ɟ": 159, + "aɜ": 160, + "ɑɨ": 161, + "i.ɜ": 162, + "eʊ": 163, + "o2": 164, + "ɐ̃": 165, + "ä": 166, + "pʲ": 167, + "kʲ": 168, + "n̩": 169, + "ɒ": 170, + "ph": 171, + "ɑu2": 172, + "uɨ": 173, + "əɪ": 174, + "ɫ": 175, + "ɬ": 176, + "yɜ": 177, + "bʲ": 178, + "ɑ2": 179, + "s̪": 180, + "aiɜ": 181, + "χ": 182, + "ɐ̃ʊ̃": 183, + "1": 184, + "ə4": 185, + "yæɜ": 186, + "a2": 187, + "ɨː": 188, + "t̪": 189, + "iouɜ": 190, + "ũ": 191, + "onɡɜ": 192, + "aɨ": 193, + "iɛ2": 194, + "ɔɨ": 195, + "ɑuɜ": 196, + "o̞": 197, + "ei2": 198, + "iou2": 199, + "c": 200, + "kː": 201, + "y2": 202, + "ɖ": 203, + "oe": 204, + "dˤ": 205, + "yɛɜ": 206, + "əʊ": 207, + "S": 208, + "ɡʲ": 209, + "onɡ2": 210, + "u\"": 211, + "eiɜ": 212, + "ʈ": 213, + "ɯᵝ": 214, + "iou5": 215, + "dZ": 216, + "r̝̊": 217, + "i.2": 218, + "tS": 219, + "s^": 220, + "ʝ": 221, + "yə5": 222, + "iɑɜ": 223, + "uə5": 224, + "pf": 225, + "ɨu": 226, + "iɑ2": 227, + "ou2": 228, + "ər2": 229, + "fʲ": 230, + "ai2": 231, + "r̝": 232, + "uəɜ": 233, + "ɳ": 234, + "əɨ": 235, + "ua5": 236, + "uɪ": 237, + "ɽ": 238, + "bː": 239, + "yu5": 240, + "uo2": 241, + "yɛ5": 242, + "l̩": 243, + "ɻ": 244, + "ərɜ": 245, + "ʂ": 246, + "i̪2": 247, + "ouɜ": 248, + "uaɜ": 249, + "a.": 250, + "a.ː": 251, + "yæ5": 252, + "dː": 253, + "r̩": 254, + "ee": 255, + "ɪu": 256, + "ər5": 257, + "i̪ɜ": 258, + "æi": 259, + "u:": 260, + "i.ː": 261, + "t^": 262, + "o1": 263, + "ɪ^": 264, + "ai": 265, + "ueiɜ": 266, + "æː": 267, + "ɛɪ": 268, + "eə": 269, + "i.": 270, + "ɴ": 271, + "ie": 272, + "ua2": 273, + "ɑ1": 274, + "o4": 275, + "tʃː": 276, + "o:": 277, + "ɑ:": 278, + "u1": 279, + "N": 280, + "i̪1": 281, + "au": 282, + "yæ2": 283, + "u.": 284, + "qː": 285, + "yəɜ": 286, + "y:": 287, + "kʰ": 288, + "tʃʰ": 289, + "iʊ": 290, + "sx": 291, + "õ": 292, + "uo": 293, + "tʰ": 294, + "uai5": 295, + "bʰ": 296, + "u.ː": 297, + "uə2": 298, + "ʊə": 299, + "d^": 300, + "s̪ː": 301, + "yiɜ": 302, + "dʰ": 303, + "r.": 304, + "oe:": 305, + "i1": 306, + "ɟː": 307, + "yu2": 308, + "nʲʲ": 309, + "i̪4": 310, + "uei2": 311, + "tsʲ": 312, + "ɸ": 313, + "ĩ": 314, + "ɑ4": 315, + "t̪ː": 316, + "eɑ": 317, + "u4": 318, + "e:": 319, + "tsː": 320, + "ʈʰ": 321, + "ɡʰ": 322, + "ɯɯ": 323, + "dʒʲ": 324, + "ʂʲ": 325, + "X": 326, + "ɵː": 327, + "uaiɜ": 328, + "tɕʲ": 329, + "ã": 330, + "t^ː": 331, + "ẽː": 332, + "yɛ2": 333, + "cː": 334, + "i.1": 335, + "ɛʊ": 336, + "dˤdˤ": 337, + "dʒː": 338, + "i4": 339, + "ɡː": 340, + "yi": 341, + "ɕʲ": 342, + "ɟʰ": 343, + "pʰ": 344, + "dʑʲ": 345, + "yuɜ": 346, + "ua1": 347, + "ua4": 348, + "æiː": 349, + "ɐɐ": 350, + "ui": 351, + "iou1": 352, + "ʊː": 353, + "a1": 354, + "iou4": 355, + "cʰ": 356, + "iɛ1": 357, + "yə2": 358, + "ɖʰ": 359, + "ẽ": 360, + "ʒʲ": 361, + "ää": 362, + "ər4": 363, + "iːː": 364, + "ɪː": 365, + "iɑ1": 366, + "ər1": 367, + "œː": 368, + "øi": 369, + "ɪuː": 370, + "cʰcʰ": 371, + "əː1": 372, + "iː1": 373, + "ũ": 374, + "kʰː": 375, + "o̞o̞": 376, + "xʲ": 377, + "ou1": 378, + "iɛ4": 379, + "e̞e̞": 380, + "y1": 381, + "dzː": 382, + "dʲʲ": 383, + "dʰː": 384, + "ɯᵝɯᵝ": 385, + "lː": 386, + "uo1": 387, + "i.4": 388, + "i:": 389, + "yɛ5ʲ": 390, + "a4": 391, + "": 509, + "": 508, + "": 507, + "": 506, + "": 505 + }, + "language": "en-us" + }, + "feature_space": { + "version": "v2", + "embedder_sha256": "d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e", + "embedder_config_sha256": "2ddde063b795d38d9051a7215a092fecf4cfe148b54251e38de51d88d356898b", + "frame_hz": 100, + "feature_dim": 768, + "pitch": "rmvpe", + "rmvpe_sha256": "6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193" + }, + "source_attribution": "LJ Speech 1.1 (Keith Ito and Linda Johnson); public-domain recordings and annotations. https://keithito.com/LJ-Speech-Dataset/", + "initialized_from_sha256": "ede4a46be5e0eddd979848313ef3af65d8836f20f59cc52c29a1e30e24489b81", + "training_manifest_sha256": "09ee901451c3ce0fdf0186a4e5b18c693f8dfc5c95b1a0945944eb041d8025c6", + "license": "CC-BY-4.0", + "license_url": "https://creativecommons.org/licenses/by/4.0/", + "quality_status": "Development checkpoint; naturalness and pitch stability still require listening acceptance", + "url": null +} diff --git a/Resources/tts/es/README.md b/Resources/tts/es/README.md new file mode 100644 index 0000000000000000000000000000000000000000..5444856303e1304e2044a69387e4750ccb4e72a7 --- /dev/null +++ b/Resources/tts/es/README.md @@ -0,0 +1,27 @@ +--- +language: +- es +license: cc-by-4.0 +tags: +- applio +- text-to-speech +--- +# Applio TTS companion (es) + +A duration-based text-to-ContentVec companion. Requires the Applio text-RVC runtime and a compatible RVC v2 voice; this is not an RVC generator/discriminator pretrain or a standalone audio decoder. + +Import `frontend.pt` using Applio's TTS model upload control, then select the same speech language. RVC voices themselves are language agnostic. This checkpoint supplies pronunciation and timing only for its stated language; additional languages are not validated by its existence. + +SHA-256: `4f8eda19e75634fe438ba232b3a6fa90f11c19ce078748da4b61a5570f4a6c99`. Checkpoint file bytes: 23197714. No source recordings, transcripts, recognizers or RVC voice weights are included. + +## Training and attribution + +Multilingual LibriSpeech (Vineel Pratap et al., Meta/Facebook), CC BY 4.0, publisher facebook/multilingual_librispeech revision 2e83e61823b4c47dcbcb1980bb88601274127609, speakers 10246 and 8881. https://huggingface.co/datasets/facebook/multilingual_librispeech . Initialization also descends from the LJ Speech public-domain base. + +Architecture, feature-space version and initialization hash are in `model.json`. The checkpoint itself retains training/validation identities and initialization provenance. Applio engine source: https://github.com/IAHispano/Applio . Checkpoint redistribution is offered under CC BY 4.0 with the attribution above: https://creativecommons.org/licenses/by/4.0/ . + +## Limitations and evaluation + +This is a development checkpoint. Intelligibility has been measured on held-out recordings through an unchanged RVC decoder, but natural rhythm and stable pitch have not received human acceptance. Final evaluation sets have already been used and must not be reused for tuning. Automatically generated transcripts can contain errors. These shared bases were trained using paired source transcripts, not automatically transcribed data. + +Current measured comparisons and playable samples: https://github.com/IAHispano/Applio-App/pull/22 . Figures for voice-adapted companions must not be attributed to this shared base. diff --git a/Resources/tts/es/frontend.pt b/Resources/tts/es/frontend.pt new file mode 100644 index 0000000000000000000000000000000000000000..7f06c1f724c06030403e54642f4b3ae56bc99f2b --- /dev/null +++ b/Resources/tts/es/frontend.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f8eda19e75634fe438ba232b3a6fa90f11c19ce078748da4b61a5570f4a6c99 +size 23197714 diff --git a/Resources/tts/es/model.json b/Resources/tts/es/model.json new file mode 100644 index 0000000000000000000000000000000000000000..de9b30450494d7cb316b0b18d94d5add91ef4393 --- /dev/null +++ b/Resources/tts/es/model.json @@ -0,0 +1,431 @@ +{ + "id": "applio-tts-es", + "language": "es", + "sha256": "4f8eda19e75634fe438ba232b3a6fa90f11c19ce078748da4b61a5570f4a6c99", + "bytes": 23197714, + "format": "applio-text-rvc-frontend-v1", + "config": { + "feature_dim": 768, + "hidden": 256, + "layers": 4, + "reduction": 1, + "architecture": "duration", + "phonetic": true, + "phoneme_vocabulary": { + "": 1, + "": 0, + "": 2, + "": 3, + "n": 4, + "s": 5, + "t": 6, + "ə": 7, + "l": 8, + "a": 9, + "i": 10, + "k": 11, + "d": 12, + "m": 13, + "ɛ": 14, + "ɾ": 15, + "e": 16, + "ɪ": 17, + "p": 18, + "o": 19, + "ɐ": 20, + "z": 21, + "ð": 22, + "f": 23, + "j": 24, + "v": 25, + "b": 26, + "ɹ": 27, + "ʁ": 28, + "ʊ": 29, + "iː": 30, + "r": 31, + "w": 32, + "ʌ": 33, + "u": 34, + "ɡ": 35, + "æ": 36, + "aɪ": 37, + "ʃ": 38, + "h": 39, + "ɔ": 40, + "ɑː": 41, + "ŋ": 42, + "ɚ": 43, + "eɪ": 44, + "β": 45, + "uː": 46, + "y": 47, + "ɑ̃": 48, + "oʊ": 49, + "ᵻ": 50, + "eː": 51, + "θ": 52, + "aʊ": 53, + "ts": 54, + "oː": 55, + "ɔ̃": 56, + "ɣ": 57, + "ɜ": 58, + "ɑ": 59, + "dʒ": 60, + "əl": 61, + "x": 62, + "ɜː": 63, + "ç": 64, + "ʒ": 65, + "tʃ": 66, + "ɔː": 67, + "ɑːɹ": 68, + "ɛ̃": 69, + "ʎ": 70, + "ɔːɹ": 71, + "ʋ": 72, + "aː": 73, + "ɕ": 74, + "œ": 75, + "ø": 76, + "oːɹ": 77, + "ɲ": 78, + "yː": 79, + "ʔ": 80, + "iə": 81, + "i5": 82, + "s.": 83, + "tɕ": 84, + "??": 85, + "nʲ": 86, + "ɛː": 87, + "œ̃": 88, + "ɭ": 89, + "ɔø": 90, + "ʑ": 91, + "tʲ": 92, + "ɨ": 93, + "ɛɹ": 94, + "ts.": 95, + "rʲ": 96, + "ɪɹ": 97, + "ɭʲ": 98, + "i.5": 99, + "ɔɪ": 100, + "q": 101, + "sʲ": 102, + "u5": 103, + "ʊɹ": 104, + "iɜ": 105, + "a5": 106, + "iɛ5": 107, + "øː": 108, + "ʕ": 109, + "ja": 110, + "əɜ": 111, + "th": 112, + "ɑ5": 113, + "oɪ": 114, + "dʲ": 115, + "ə5": 116, + "tɕh": 117, + "ts.h": 118, + "mʲ": 119, + "ɯ": 120, + "dʑ": 121, + "vʲ": 122, + "e̞": 123, + "tʃʲ": 124, + "ei5": 125, + "o5": 126, + "onɡ5": 127, + "ɑu5": 128, + "iɑ5": 129, + "ai5": 130, + "aɪɚ": 131, + "kh": 132, + "ə1": 133, + "ʐ": 134, + "i2": 135, + "ʉ": 136, + "ħ": 137, + "t[": 138, + "aɪə": 139, + "ʲ": 140, + "ju": 141, + "ə2": 142, + "u2": 143, + "oɜ": 144, + "pː": 145, + "iɛɜ": 146, + "ou5": 147, + "y5": 148, + "uɜ": 149, + "tː": 150, + "uo5": 151, + "d[": 152, + "uoɜ": 153, + "tsh": 154, + "ɑɜ": 155, + "ɵ": 156, + "i̪5": 157, + "uei5": 158, + "ɟ": 159, + "aɜ": 160, + "ɑɨ": 161, + "i.ɜ": 162, + "eʊ": 163, + "o2": 164, + "ɐ̃": 165, + "ä": 166, + "pʲ": 167, + "kʲ": 168, + "n̩": 169, + "ɒ": 170, + "ph": 171, + "ɑu2": 172, + "uɨ": 173, + "əɪ": 174, + "ɫ": 175, + "ɬ": 176, + "yɜ": 177, + "bʲ": 178, + "ɑ2": 179, + "s̪": 180, + "aiɜ": 181, + "χ": 182, + "ɐ̃ʊ̃": 183, + "1": 184, + "ə4": 185, + "yæɜ": 186, + "a2": 187, + "ɨː": 188, + "t̪": 189, + "iouɜ": 190, + "ũ": 191, + "onɡɜ": 192, + "aɨ": 193, + "iɛ2": 194, + "ɔɨ": 195, + "ɑuɜ": 196, + "o̞": 197, + "ei2": 198, + "iou2": 199, + "c": 200, + "kː": 201, + "y2": 202, + "ɖ": 203, + "oe": 204, + "dˤ": 205, + "yɛɜ": 206, + "əʊ": 207, + "S": 208, + "ɡʲ": 209, + "onɡ2": 210, + "u\"": 211, + "eiɜ": 212, + "ʈ": 213, + "ɯᵝ": 214, + "iou5": 215, + "dZ": 216, + "r̝̊": 217, + "i.2": 218, + "tS": 219, + "s^": 220, + "ʝ": 221, + "yə5": 222, + "iɑɜ": 223, + "uə5": 224, + "pf": 225, + "ɨu": 226, + "iɑ2": 227, + "ou2": 228, + "ər2": 229, + "fʲ": 230, + "ai2": 231, + "r̝": 232, + "uəɜ": 233, + "ɳ": 234, + "əɨ": 235, + "ua5": 236, + "uɪ": 237, + "ɽ": 238, + "bː": 239, + "yu5": 240, + "uo2": 241, + "yɛ5": 242, + "l̩": 243, + "ɻ": 244, + "ərɜ": 245, + "ʂ": 246, + "i̪2": 247, + "ouɜ": 248, + "uaɜ": 249, + "a.": 250, + "a.ː": 251, + "yæ5": 252, + "dː": 253, + "r̩": 254, + "ee": 255, + "ɪu": 256, + "ər5": 257, + "i̪ɜ": 258, + "æi": 259, + "u:": 260, + "i.ː": 261, + "t^": 262, + "o1": 263, + "ɪ^": 264, + "ai": 265, + "ueiɜ": 266, + "æː": 267, + "ɛɪ": 268, + "eə": 269, + "i.": 270, + "ɴ": 271, + "ie": 272, + "ua2": 273, + "ɑ1": 274, + "o4": 275, + "tʃː": 276, + "o:": 277, + "ɑ:": 278, + "u1": 279, + "N": 280, + "i̪1": 281, + "au": 282, + "yæ2": 283, + "u.": 284, + "qː": 285, + "yəɜ": 286, + "y:": 287, + "kʰ": 288, + "tʃʰ": 289, + "iʊ": 290, + "sx": 291, + "õ": 292, + "uo": 293, + "tʰ": 294, + "uai5": 295, + "bʰ": 296, + "u.ː": 297, + "uə2": 298, + "ʊə": 299, + "d^": 300, + "s̪ː": 301, + "yiɜ": 302, + "dʰ": 303, + "r.": 304, + "oe:": 305, + "i1": 306, + "ɟː": 307, + "yu2": 308, + "nʲʲ": 309, + "i̪4": 310, + "uei2": 311, + "tsʲ": 312, + "ɸ": 313, + "ĩ": 314, + "ɑ4": 315, + "t̪ː": 316, + "eɑ": 317, + "u4": 318, + "e:": 319, + "tsː": 320, + "ʈʰ": 321, + "ɡʰ": 322, + "ɯɯ": 323, + "dʒʲ": 324, + "ʂʲ": 325, + "X": 326, + "ɵː": 327, + "uaiɜ": 328, + "tɕʲ": 329, + "ã": 330, + "t^ː": 331, + "ẽː": 332, + "yɛ2": 333, + "cː": 334, + "i.1": 335, + "ɛʊ": 336, + "dˤdˤ": 337, + "dʒː": 338, + "i4": 339, + "ɡː": 340, + "yi": 341, + "ɕʲ": 342, + "ɟʰ": 343, + "pʰ": 344, + "dʑʲ": 345, + "yuɜ": 346, + "ua1": 347, + "ua4": 348, + "æiː": 349, + "ɐɐ": 350, + "ui": 351, + "iou1": 352, + "ʊː": 353, + "a1": 354, + "iou4": 355, + "cʰ": 356, + "iɛ1": 357, + "yə2": 358, + "ɖʰ": 359, + "ẽ": 360, + "ʒʲ": 361, + "ää": 362, + "ər4": 363, + "iːː": 364, + "ɪː": 365, + "iɑ1": 366, + "ər1": 367, + "œː": 368, + "øi": 369, + "ɪuː": 370, + "cʰcʰ": 371, + "əː1": 372, + "iː1": 373, + "ũ": 374, + "kʰː": 375, + "o̞o̞": 376, + "xʲ": 377, + "ou1": 378, + "iɛ4": 379, + "e̞e̞": 380, + "y1": 381, + "dzː": 382, + "dʲʲ": 383, + "dʰː": 384, + "ɯᵝɯᵝ": 385, + "lː": 386, + "uo1": 387, + "i.4": 388, + "i:": 389, + "yɛ5ʲ": 390, + "a4": 391, + "": 509, + "": 508, + "": 507, + "": 506, + "": 505 + }, + "language": "es" + }, + "feature_space": { + "version": "v2", + "embedder_sha256": "d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e", + "embedder_config_sha256": "2ddde063b795d38d9051a7215a092fecf4cfe148b54251e38de51d88d356898b", + "frame_hz": 100, + "feature_dim": 768, + "pitch": "rmvpe", + "rmvpe_sha256": "6d62215f4306e3ca278246188607209f09af3dc77ed4232efdd069798c4ec193" + }, + "source_attribution": "Multilingual LibriSpeech (Vineel Pratap et al., Meta/Facebook), CC BY 4.0, publisher facebook/multilingual_librispeech revision 2e83e61823b4c47dcbcb1980bb88601274127609, speakers 10246 and 8881. https://huggingface.co/datasets/facebook/multilingual_librispeech . Initialization also descends from the LJ Speech public-domain base.", + "initialized_from_sha256": "bd6b4136d04c0b80cf8d5b790d0c54d6e8a374758c2abd24067070a59647b10a", + "training_manifest_sha256": "4a6567d78f87e36f1c79f8c6296c436bd6238a2c50df7e13ffe10e6f7e05c9d6", + "license": "CC-BY-4.0", + "license_url": "https://creativecommons.org/licenses/by/4.0/", + "quality_status": "Development checkpoint; naturalness and pitch stability still require listening acceptance", + "url": null +} diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..79a65417f6a7379c4370d5123141e981d8d4ebdc --- /dev/null +++ b/config.json @@ -0,0 +1,3 @@ +{ + "arch_version": "applio" +} \ No newline at end of file diff --git a/env.zip b/env.zip new file mode 100644 index 0000000000000000000000000000000000000000..d60fb4c9450280f0d0805c14fb04f13f58dd3889 --- /dev/null +++ b/env.zip @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e95c521075e4309f192739594b5de0bba41dfbb8a17cc013c4a1a975a7f76802 +size 21086039 diff --git a/pretrains.json b/pretrains.json new file mode 100644 index 0000000000000000000000000000000000000000..3cb78f4a5abec9688b8af00e0ed3189e935cb86d --- /dev/null +++ b/pretrains.json @@ -0,0 +1,120 @@ +{ + "Beatz Forge": { + "32k": { + "D": "SeoulStreamingStation/KLM_BEATMASTER/resolve/main/D_BeatzForge_V2_32k.pth", + "G": "SeoulStreamingStation/KLM_BEATMASTER/resolve/main/G_BeatzForge_V2_32k.pth" + } + }, + "DMR V1": { + "32k": { + "D": "Razer112/DMR_Pretrain/resolve/main/D_DMR-V1.pth", + "G": "Razer112/DMR_Pretrain/resolve/main/G_DMR-V1.pth" + } + }, + "Italia": { + "32k": { + "D": "TheStinger/itaila/resolve/main/ItaIla_32k_D.pth", + "G": "TheStinger/itaila/resolve/main/ItaIla_32k_G.pth" + } + }, + "IMA V1": { + "32k": { + "D": "Loren85/IMA-TEST-V1/resolve/main/D_2333333.pth", + "G": "Loren85/IMA-TEST-V1/resolve/main/G_2333333.pth" + } + }, + "KLM4": { + "32k": { + "D": "SeoulStreamingStation/KLM49_HFG/resolve/main/D_KLM_HFG_32k.pth", + "G": "SeoulStreamingStation/KLM49_HFG/resolve/main/G_KLM_HFG_32k.pth" + }, + "40k": { + "D": "SeoulStreamingStation/KLM49_HFG/resolve/main/D_KLM_HFG_40k.pth", + "G": "SeoulStreamingStation/KLM49_HFG/resolve/main/G_KLM_HFG_40k.pth" + }, + "48k": { + "D": "SeoulStreamingStation/KLM49_HFG/resolve/main/D_KLM_HFG_48k.pth", + "G": "SeoulStreamingStation/KLM49_HFG/resolve/main/G_KLM_HFG_48k.pth" + } + }, + "Nanashi v1.7": { + "32k": { + "D": "shiromiya/nanashi-pretrain/resolve/main/v1.7/D_nanashi_v1_7.pth", + "G": "shiromiya/nanashi-pretrain/resolve/main/v1.7/G_nanashi_v1_7.pth" + } + }, + "Ov2Super": { + "32k": { + "D": "ORVC/Ov2Super/resolve/main/f0Ov2Super32kD.pth", + "G": "ORVC/Ov2Super/resolve/main/f0Ov2Super32kG.pth" + }, + "40k": { + "D": "ORVC/Ov2Super/resolve/main/f0Ov2Super40kD.pth", + "G": "ORVC/Ov2Super/resolve/main/f0Ov2Super40kG.pth" + } + }, + "RIN_E3": { + "40k": { + "D": "MUSTAR/RIN_E3/resolve/main/RIN_E3_D.pth", + "G": "MUSTAR/RIN_E3/resolve/main/RIN_E3_G.pth" + } + }, + "Rigel Base": { + "32k": { + "D": "MUSTAR/Rigel-rvc-base-pretrained-model/resolve/main/Rigel_32k_Base_and_FineTuned/Base-model_32k_fp32/D_Rigel_32k_3890220.pth", + "G": "MUSTAR/Rigel-rvc-base-pretrained-model/resolve/main/Rigel_32k_Base_and_FineTuned/Base-model_32k_fp32/G_Rigel_32k_3890220.pth" + } + }, + "Rigel FineTuned": { + "32k": { + "D": "MUSTAR/Rigel-rvc-base-pretrained-model/resolve/main/Rigel_32k_Base_and_FineTuned/FineTuned-model_32k_fp32/D_Rigel_32k_fp32_2854856.pth", + "G": "MUSTAR/Rigel-rvc-base-pretrained-model/resolve/main/Rigel_32k_Base_and_FineTuned/FineTuned-model_32k_fp32/G_Rigel_32k_fp32_2854856.pth" + } + }, + "Snowie": { + "32k": { + "D": "MUSTAR/SnowieV3.1-32k/resolve/main/D_SnowieV3.1_32k.pth", + "G": "MUSTAR/SnowieV3.1-32k/resolve/main/G_SnowieV3.1_32k.pth" + }, + "40k": { + "D": "MUSTAR/SnowieV3.1-40k/resolve/main/D_SnowieV3.1_40k.pth", + "G": "MUSTAR/SnowieV3.1-40k/resolve/main/G_SnowieV3.1_40k.pth" + }, + "48k": { + "D": "MUSTAR/SnowieV3.1-48k/resolve/main/D_SnowieV3.1_48k.pth", + "G": "MUSTAR/SnowieV3.1-48k/resolve/main/G_SnowieV3.1_48k.pth" + } + }, + "Snowie-X-RinE3": { + "40k": { + "D": "MUSTAR/SnowieV3.1-X-RinE3-40K/resolve/main/D_Snowie-X-Rin_40k.pth", + "G": "MUSTAR/SnowieV3.1-X-RinE3-40K/resolve/main/G_Snowie-X-Rin_40k.pth" + } + }, + "SingerPreTrain": { + "32k": { + "D": "Sztef/SingerPreTrained/resolve/main/update/f0D_SingerPreTrain.pth", + "G": "Sztef/SingerPreTrained/resolve/main/update/f0G_SingerPreTrain.pth" + } + }, + "Titan": { + "32k": { + "D": "blaise-tk/TITAN/resolve/main/models/medium/32k/pretrained/D-f032k-TITAN-Medium.pth", + "G": "blaise-tk/TITAN/resolve/main/models/medium/32k/pretrained/G-f032k-TITAN-Medium.pth" + }, + "40k": { + "D": "blaise-tk/TITAN/resolve/main/models/medium/40k/pretrained/D-f040k-TITAN-Medium.pth", + "G": "blaise-tk/TITAN/resolve/main/models/medium/40k/pretrained/G-f040k-TITAN-Medium.pth" + }, + "48k": { + "D": "blaise-tk/TITAN/resolve/main/models/medium/48k/pretrained/D-f048k-TITAN-Medium.pth", + "G": "blaise-tk/TITAN/resolve/main/models/medium/48k/pretrained/G-f048k-TITAN-Medium.pth" + } + }, + "UKA": { + "32k": { + "D": "Plasmati/Pretrains/resolve/main/UKA-Pretrain-D.pth", + "G": "Plasmati/Pretrains/resolve/main/UKA-Pretrain-G.pth" + } + } +} \ No newline at end of file