Third Party Notices
Release date: November 11, 2026
Third-party components remain subject to their own licenses. The proprietary Marks Dictation license does not restrict rights those licenses grant. Provider names do not imply endorsement.
1 Speech models
NVIDIA Parakeet TDT 0.6B v2 and v3, by NVIDIA Corporation, are licensed under Creative Commons Attribution 4.0 International. Sources: Parakeet v2 and Parakeet v3.
The models were converted to ONNX and quantized to int8 by the sherpa-onnx project. Further release changes: none to the model weights; the app adds a word-piece list (bpe.vocab) generated from each model's own token list so that saved vocabulary can guide recognition. Conversion source, revision, and retained contributor notices: the sherpa-onnx "asr-models" release (https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models), files sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8 and sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8, converted from NVIDIA's Parakeet TDT 0.6B v2 and v3 models and downloaded with SHA-256 verification; the CC BY 4.0 attribution above is retained.
2 Software
sherpa-onnx uses Apache License 2.0. ONNX Runtime, Silero VAD, Velopack, .NET Runtime, and Windows Presentation Foundation have MIT license texts in the licenses directory.
Release versions, copyright statements, modifications, additional dependencies, and required NOTICE files: sherpa-onnx 1.13.8 (Apache License 2.0, copyright the k2-fsa/sherpa-onnx authors); ONNX Runtime 1.28.2 (MIT License, copyright Microsoft Corporation); Silero VAD model silero_vad.onnx from the sherpa-onnx asr-models release (MIT License, copyright Silero Team); Velopack 1.2.161 (MIT License); .NET Runtime and Windows Desktop 10.0.9 (MIT License, copyright .NET Foundation and contributors). These are the versions in Marks Dictation 0.1.7 and must be rechecked for each release. Applicable supplemental redistribution terms accompany their components. Preserve the included license and attribution texts; their warranty provisions apply subject to law.