Terms of use

These terms govern your use of the edytlab desktop application and this marketing website. By installing or using the app you agree to these terms. They are short on purpose.

Licence

edytlab is released under the MIT licence; the LICENSE file in the repository is the authoritative version. You may use, copy, and modify the source for any purpose, subject to retaining the copyright notice.

Third-party services

When you connect an LLM provider (Anthropic, OpenAI, Google Gemini, Groq, OpenRouter, or a local model) you are subject to that provider's terms. The same goes for the ChatGPT plan: you sign in to OpenAI, and OpenAI's terms and the terms of your ChatGPT plan apply to the usage, which counts toward that plan. edytlab is not a party to that relationship and is not responsible for charges you incur with that provider.

The speech model you can download from Settings is Whisper base, which OpenAI publishes under the Apache-2.0 licence (its code is MIT). edytlab does not include or redistribute it: when you click Download, the app fetches the files from Hugging Face, whose terms apply to that download, and they are kept on your computer. See the model card and the privacy page.

The speaker model you can download from Settings is two files. The speaker-turn (segmentation) network is pyannote's segmentation-3.0, released under the MIT licence (© CNRS) and downloaded as the ONNX export onnx-community/pyannote-segmentation-3.0. The voiceprint (speaker embedding) network is WeSpeaker ResNet34-LM, by the WeSpeaker authors, licensed under Creative Commons Attribution 4.0 (CC BY 4.0). edytlab does not include or redistribute either: when you click Download, the app fetches the files from Hugging Face, whose terms apply to that download, and they are kept on your computer.

The stem model you can download from Settings is two files, the graph and its weights, of Meta's HT-Demucs v4 (Demucs is released under the MIT licence, © Meta Platforms, Inc. and affiliates), downloaded as the ONNX export Intel/demucs-openvino by Intel, which is also marked MIT. edytlab does not include or redistribute it: when you click Download, the app fetches the files from Hugging Face, whose terms apply to that download, and they are kept on your computer. The model was trained on the MUSDB18-HQ dataset and about 800 further songs; the licence of the weights is the one the authors state, MIT.

Third-party licences

edytlab is built from open source libraries, fonts and code, each under its own licence. Their licence texts and copyright notices ship with every installer, are shown in the app under Settings, Account, Open source licences, and are in the repository as THIRD_PARTY_NOTICES.txt. The app's Geist and Geist Mono fonts (© 2024 The Geist Project Authors) and this website's Inter font (© 2016 The Inter Project Authors, from its licence) are under the SIL Open Font License 1.1.

Disclaimer

The software is provided “as is”, without warranty of any kind, express or implied. We do not warrant that audio output will match any particular reference, that ML models will produce specific results, or that the app is free of defects.

Limitation of liability

To the fullest extent permitted by law, the authors and contributors are not liable for any indirect, incidental, special, or consequential damages arising out of your use of the software, including loss of audio data. Keep backups.