| --- |
| license: mit |
| tags: |
| - audio |
| - music |
| - stem-separation |
| - vocal-removal |
| - onnx |
| - mdx-net |
| - uvr |
| - demucs |
| language: |
| - en |
| --- |
| |
| # UVR5 & Demucs Models — Backup Repository |
|
|
| This repository is a **backup and redistribution mirror** of select pre-trained ONNX models from the [Ultimate Vocal Remover (UVR)](https://github.com/Anjok07/ultimatevocalremovergui) project and the [Demucs](https://github.com/facebookresearch/demucs) project by Facebook Research. The models are hosted here unaltered for reliability and availability purposes. This repository is not affiliated with the original authors. |
|
|
| ## Purpose |
|
|
| The canonical source for these models is the [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo) GitHub release archive. This mirror exists to provide a stable, versioned download location independent of GitHub release availability. |
|
|
| Redistribution is permitted under the [MIT License](https://opensource.org/licenses/MIT), which explicitly grants the right to use, copy, and distribute the models provided that the original copyright notice and license text are preserved. Each model in this repository is accompanied by a `.LICENSE` file containing the full attribution and license terms. |
|
|
| ## Models |
|
|
| | File | Size | Description | Original Source | |
| |---|---|---|---| |
| | `UVR-MDX-NET-Inst_HQ_3.onnx` | 64 MB | High-quality instrumental/vocal separation | [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo/releases/download/all_public_uvr_models/UVR-MDX-NET-Inst_HQ_3.onnx) | |
| | `UVR_MDXNET_Main.onnx` | 64 MB | Main MDX-Net vocal separation model | [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo/releases/download/all_public_uvr_models/UVR_MDXNET_Main.onnx) | |
| | `UVR_MDXNET_KARA_2.onnx` | 50 MB | Karaoke stem separation (v2) | [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo/releases/download/all_public_uvr_models/UVR_MDXNET_KARA_2.onnx) | |
| | `UVR_MDXNET_KARA.onnx` | 28 MB | Karaoke stem separation | [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo/releases/download/all_public_uvr_models/UVR_MDXNET_KARA.onnx) | |
| | `UVR_MDXNET_3_9662.onnx` | 28 MB | MDX-Net v3 (96.62 SDR) | [TRvlvr/model_repo](https://github.com/TRvlvr/model_repo/releases/download/all_public_uvr_models/UVR_MDXNET_3_9662.onnx) | |
| | `htdemucs_embedded.onnx` | 172 MB | HTDemucs 4-stem source separation (vocals, drums, bass, other) — ONNX export | [timcsy/demucs-web-onnx](https://huggingface.co/timcsy/demucs-web-onnx) | |
|
|
| ## Architecture |
|
|
| All models are exported in ONNX format for cross-platform inference. |
|
|
| **KUIELab-MDX-Net** (UVR models) |
| - **Paper:** [KUIELab-MDX-Net: A Two-Stream Neural Network for Music Demixing](https://arxiv.org/abs/2111.12203) |
| - **Architecture authors:** Woosung Choi, Minseok Kim, Jaehwa Chung, Daewon Lee, Soonyoung Jung (KUIELab) |
|
|
| **Hybrid Transformer Demucs (HTDemucs)** |
| - **Paper:** [Hybrid Transformers for Music Source Separation](https://arxiv.org/abs/2211.00847) |
| - **Architecture authors:** Alexandre Défossez, Nicolas Usunier, Léon Bottou, Francis Bach (Facebook Research / Meta AI) |
|
|
| ## Credits |
|
|
| **Ultimate Vocal Remover (UVR) models:** |
| - **UVR Project:** [Anjok07 (DilanBoskan)](https://github.com/Anjok07), KimberleyJSN, and the UVR development community |
| - **UVR Repository:** https://github.com/Anjok07/ultimatevocalremovergui |
| - **Model Distribution:** https://github.com/TRvlvr/model_repo |
| |
| **HTDemucs model:** |
| - **Demucs Project:** Alexandre Défossez and contributors (Facebook Research / Meta AI) |
| - **Demucs Repository:** https://github.com/facebookresearch/demucs |
| - **ONNX Conversion:** [timcsy/demucs-web-onnx](https://huggingface.co/timcsy/demucs-web-onnx) |
| |
| If you use these models in your work, please credit the respective original authors. |
| |
| ## License |
| |
| MIT — see the `.LICENSE` file accompanying each model for the full license text and attribution notice. |
| |
