Skip to content

Notices

Open-source and model notices

SonicStem is built with open-source software and pretrained audio models created by others. This page acknowledges them and reproduces the notices that applicable licences ask us to include. Where a licence is named below, it applies to the part named. Listing a component here does not mean its authors endorse SonicStem.

Last updated 6 October 2026

Software delivered to your browser

The SonicStem web app sends the following components to your browser as part of the app.

  • React, React DOM and Scheduler 19.2.7, 19.2.7 and 0.27.0 — MIT License — Copyright (c) Meta Platforms, Inc. and affiliates.
  • signalsmith-stretch 1.3.2, with SonicStem modifications — MIT License — Copyright (c) 2022 Geraint Luff / Signalsmith Audio Ltd.
  • web-audio-beat-detector 8.2.36, web-audio-beat-detector-broker 5.0.14, web-audio-beat-detector-worker 6.0.13, worker-factory 7.0.49, automation-events 7.1.19, broker-factory 3.1.14 and fast-unique-numbers 9.0.27 — MIT License — Copyright (c) 2026 Christoph Guttandin.
  • standardized-audio-context 25.3.77 — MIT License — Copyright (c) 2024 Christoph Guttandin.
  • libFLAC 1.5.0, compiled to WebAssembly — BSD-3-Clause — Copyright (C) 2000-2009 Josh Coalson; Copyright (C) 2011-2025 Xiph.Org Foundation. The licence is reproduced below.
  • Emscripten runtime, the JavaScript that loads the WebAssembly builds above — MIT License — Copyright (c) 2010-2014 Emscripten authors.
  • musl libc, portions of which may be compiled into those WebAssembly builds — MIT License — Copyright © 2005-2020 Rich Felker, et al.
  • Vite module-preload helper — MIT License — Copyright (c) 2019-present, VoidZero Inc. and Vite contributors.

The MIT License text that applies to each MIT component, with that component's copyright line, is reproduced below.

Audio models used by the tools

These models run on computers we operate. They are not sent to your browser.

  • All Stems: htdemucs_ft, a pretrained Hybrid Transformer Demucs model released by the Demucs authors at Meta. Demucs code: Copyright (c) Meta Platforms, Inc. and affiliates, MIT License.
  • Vocals / Music: Mel-Band RoFormer vocal model by Kimberley Jensen. Model: MIT License, as published on its model card. Inference code: adapted from Music Source Separation Training, Copyright (c) 2024 Roman Solovyev (ZFTurbo), MIT License, which includes RoFormer architecture code by Phil Wang, Copyright (c) 2023 Phil Wang, MIT License.
  • Key and Camelot: S-KEY by Deezer Research. S-KEY code: Copyright (c) 2019-present, Deezer SA, MIT License.
  • MIDI: Basic Pitch by Spotify. Code and model: Copyright 2022 Spotify AB, Apache License 2.0.
  • Lyrics: Whisper large-v3 by OpenAI, in the CTranslate2 conversion published by SYSTRAN, run with faster-whisper.
    • Model: Copyright (c) 2022 OpenAI. OpenAI's Whisper repository states that Whisper's code and model weights are released under the MIT License. The Hugging Face model card for whisper-large-v3 lists the Apache License 2.0.
    • Conversion: published by SYSTRAN under the MIT License on its model card.
    • Code: faster-whisper, Copyright (c) 2023 SYSTRAN, MIT License; CTranslate2, Copyright (c) 2018- SYSTRAN, MIT License.

Audio-analysis and server software

These components run on our servers and GPU hardware. They are not distributed to you.

  • Essentia 2.1b6.dev1389 — Universitat Pompeu Fabra — GNU Affero General Public License v3.0. SonicStem runs an unmodified copy of the published Essentia package for tempo (BPM) detection. That package includes FFTW 3.3.2 (GNU GPL), libsamplerate 0.1.8 (GNU GPL) and FFmpeg 2.8.12 (GNU LGPL). Essentia's source code is available from the Essentia project at github.com/MTG/essentia.
  • librosa — librosa development team — ISC License. Used for audio analysis, including as a tempo fallback.
  • NumPy, SciPy, scikit-learn — BSD-3-Clause.
  • Numba (Anaconda, Inc.) and llvmlite (Continuum Analytics, Inc.) — BSD-2-Clause. llvmlite includes LLVM, which is under the Apache License 2.0 with LLVM Exceptions.
  • python-soxr / libsoxr — GNU LGPL v2.1 or later.
  • soundfile — BSD-3-Clause, with libsndfile — GNU LGPL v2.1.
  • PyTorch — BSD-3-Clause. torchaudio — BSD-2-Clause. nnAudio and einops — MIT. tqdm — MPL-2.0 and MIT.
  • rotary-embedding-torch — Copyright (c) 2021 Phil Wang — MIT License.
  • FFmpeg — GNU LGPL builds on our servers; a GNU GPL build on our GPU hardware. FFmpeg is a trademark of Fabrice Bellard. lameenc — GNU LGPL v3.
  • Node.js packages — Express, better-sqlite3, the Stripe Node.js library and Multer (MIT); SQLite (public domain); the AWS SDK for JavaScript (Apache License 2.0).

This page acknowledges the software and models we use. It is not a statement about how any licence's conditions apply.

Licence texts

MIT License

Applies to each MIT component above, together with that component's copyright line.

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.

libFLAC — BSD-3-Clause

Reproduced from the libFLAC distribution.

Copyright (C) 2000-2009  Josh Coalson
Copyright (C) 2011-2025  Xiph.Org Foundation

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:

- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.

- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.

- Neither the name of the Xiph.Org Foundation nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Other licences

The full texts of the other licences named on this page are published here:

Questions about these notices: [email protected].

This page describes how SonicStem currently operates, written in plain language.