juunote
Licences
Last updated: 2 October 2026 · preliminary overview of the main components; the complete list of all packages will follow.
juunote is built on open-source software and freely licensed models. Thanks to everyone who works on them. The licence texts are available from the respective projects.
Software
| Component | Purpose | Licence |
|---|---|---|
| whisper.cpp / ggml (Georgi Gerganov et al.) | Speech recognition | MIT |
| sherpa-onnx (k2-fsa) | Speaker separation | Apache 2.0 |
| ONNX Runtime (Microsoft) | Model runtime | MIT |
| Tauri | App framework for Mac and Windows | MIT / Apache 2.0 |
| React | User interface | MIT |
| SQLCipher (Zetetic) / SQLite | Encrypted database | BSD licence / public domain |
| Rust crates (incl. tokio, reqwest, serde, rusqlite) | Core foundation | mostly MIT / Apache 2.0 |
Models
| Model | Purpose | Licence |
|---|---|---|
| OpenAI Whisper large-v3-turbo (ggml version) | Speech recognition | MIT |
| German fine-tune of Whisper large-v3-turbo | German speech recognition (optional) | as stated in the publisher’s model card |
| pyannote segmentation 3.0 | Speaker separation | MIT |
| 3D-Speaker ERes2Net | Voice features | Apache 2.0 |
The app downloads models only when needed, from the publishers’ servers, and verifies every file by its checksum.