Licences
Echo itself is released under the MIT licence.
Dependencies
Rust dependencies are checked by cargo deny check (see deny.toml) on every CI run. Only permissive licences are allowed: MIT, Apache-2.0 (with or without the LLVM exception), BSD-2-Clause, BSD-3-Clause, ISC, Zlib, Unicode-3.0, CC0-1.0, BSL-1.0, MPL-2.0 and CDLA-Permissive-2.0. A crate offered under several licences passes if one of them is on that list.
The runtime JavaScript dependencies:
| Package | Licence | Used for |
|---|---|---|
@huggingface/transformers | Apache-2.0 | Running Whisper and the speaker models |
onnxruntime-web | MIT | The in-browser model runtime |
onnxruntime-node | MIT | The native runtime for Whisper Large V3 |
@scure/bip39 | MIT | The recovery phrase word list |
Development-only packages (build tools, linters, test runners) are not shipped with Echo. CI runs pnpm audit --prod on every push.
Fonts and icons
| Asset | Licence |
|---|---|
| Mulish typeface | SIL Open Font License 1.1 |
| Lucide icons | ISC |
Speech models
Echo does not include any model files. You download them from Hugging Face when you choose to, and they are covered by their own licences:
| Model | Source | Licence |
|---|---|---|
| Whisper Large V3 Turbo | OpenAI, ONNX export by onnx-community | MIT |
| Whisper Large V3 | OpenAI, q8 ONNX export Xenova/whisper-large-v3 | Apache-2.0 |
| PyAnnote segmentation 3.0 | pyannote, ONNX export onnx-community/pyannote-segmentation-3.0 | MIT |
| WavLM Base Plus SV | Microsoft, ONNX export Xenova/wavlm-base-plus-sv | From Microsoft’s unilm project, which is MIT; the model card itself states no licence |
Ollama models you pull are covered by their publishers’ licences; check each model’s page on ollama.com.