VOCAL TOOLS v40 — THIRD PARTY NOTICES (reviewed 2026-11-05) Prototype engine: @audio/neural-separate 0.7.0 by audiojs — MIT Source: https://github.com/audiojs/neural Original weights: Mel-Band RoFormer by Kimberley Jensen — repository marked MIT Source: https://huggingface.co/KimberleyJSN/melbandroformer Converted weights: audiojs/mel-roformer — MIT Source: https://huggingface.co/audiojs/mel-roformer RoFormer architecture: Wang, Lu, Won, Chen, Song (ISMIR 2023) Source: https://arxiv.org/abs/2310.01809 This application does NOT bundle the 464 MB model file. The model is fetched from Hugging Face by the runtime in the user's browser on first use. The source repository does not document all rights to the training recordings. License notices do not extinguish third-party claims. MIT permits commercial use subject to retaining its notices. This is not a guarantee of third-party rights or approval by any advertising platform. MIT LICENSE TEXT (for MIT-licensed material) Copyright (c) Kimberley Jensen (audiojs/mel-roformer LICENSE) Copyright (c) Dmitry Iv (audiojs/neural LICENSE) Copyright (c) Microsoft Corporation (ONNX Runtime LICENSE) Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. UPSTREAM PACKAGE NOTICE — @audio/neural-separate 0.7.0 Source: https://cdn.jsdelivr.net/npm/@audio/neural-separate@0.7.0/NOTICE @audio/neural-separate Copyright 2026 Dmitry Iv. and audiojs contributors This product includes a JavaScript translation of parts of open-unmix-pytorch (https://github.com/sigsep/open-unmix-pytorch) and norbert (https://github.com/sigsep/norbert): filtering.py's multichannel Wiener filter and its expectation-maximization, with their defaults, and model.Separator's 300-frame windowing, into separate.js. Both under the MIT License: Copyright (c) 2019 Inria (Fabian-Robert Stöter, Antoine Liutkus) This product includes a JavaScript translation of parts of Demucs (https://github.com/facebookresearch/demucs): apply.apply_model's segment overlap-add and htdemucs.py's spectrogram framing, into separate.js. Under the MIT License: Copyright (c) Meta Platforms, Inc. and affiliates. This product includes a JavaScript translation of parts of Music-Source-Separation-Training (https://github.com/ZFTurbo/Music-Source-Separation-Training): utils/model_utils.py's demix(), its segments, fades and reflection, into separate.js. Under the MIT License: Copyright (c) 2024 Roman Solovyev (ZFTurbo) This product includes a JavaScript translation of parts of cocktail-fork-separation (https://github.com/merlresearch/cocktail-fork-separation): mrx.py's three STFTs, masks and iSTFTs, and separate.py's input loudness, into separate.js. Under the MIT License: Copyright (c) 2023 Mitsubishi Electric Research Laboratories (MERL) This product includes a JavaScript translation of parts of TIGER (https://github.com/JusperLee/TIGER): look2hear/models/tiger_dnr.py's TIGERDNR.wav_chunk_inference, its segments and their sum, into separate.js. Under the MIT License: Copyright (c) 2026 Kai Li This product includes a JavaScript translation of parts of pyloudnorm (https://github.com/csteinmetz1/pyloudnorm): meter.py's integrated loudness and iirfilter.py's K-weighting coefficients, into separate.js. Under the MIT License: Copyright (c) 2018 Christian Steinmetz The MIT License, for each notice above: Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. No model weights are included. The export scripts write them to the local cache, and the compact files of SCNet-large, SCNet, MRX and TIGER are fetched from huggingface.co/audiojs (scnet-large, scnet, mrx, tiger-dnr, each with its licence and model card); their terms are their authors' (Open-Unmix umxhq: MIT; SCNet and SCNet-large: MIT, Copyright (c) 2024 starrytong, redistributable converted with attribution, github.com/starrytong/SCNet/issues/35; Demucs htdemucs weights: see the README; MRX: MIT, Copyright 2023 Mitsubishi Electric Research Laboratories (MERL), the repository's .reuse/dep5 naming each checkpoint; TIGER's DnR model: Apache 2.0, its model card, huggingface.co/JusperLee/TIGER-DnR). PINNED MODEL PROVENANCE Original checkpoint MIT declaration: ac9b0614ab3cd7f77219e18ba494dfd93956c348 https://huggingface.co/KimberleyJSN/melbandroformer/commit/ac9b0614ab3cd7f77219e18ba494dfd93956c348 Converted model revision used by neural-separate 0.7.0: c1aba02c00a0bee852d6754fe6ede8164f90a4b3 https://huggingface.co/audiojs/mel-roformer/blob/c1aba02c00a0bee852d6754fe6ede8164f90a4b3/LICENSE ONNX SHA-256: adcbbb826f12453fbb86652b503242d565a9c81cdef1bca2fe0e54752eba91ef Runtime dependency versions resolved by the external CDN should be audited before self-hosting or bundling.