Credits
Last updated 14 November 2026
Stentor is built on AI models and tools that their makers share openly. Thank you to them. Each is listed here with its licence; every one of them allows use in a paid app. The same list is in Stentor, in Settings › About › Credits.
AI models
Downloaded by Stentor from Hugging Face when you choose them, and run on your computer.
- Chatterbox Turbo, Standard and Multilingual by Resemble AI · MIT. Your voice, moods and sounds, and 23 languages.
- Resemble Enhance by Resemble AI · MIT. Cleaning up recordings.
- Whisper large-v3-turbo by OpenAI (MLX version by the MLX Community) · MIT. Word check and caption timings.
- Qwen3-TTS by Qwen, Alibaba Cloud (MLX version by the MLX Community) · Apache 2.0. Fast English voiceovers, and designing voices.
- Qwen3 4B Instruct by Qwen, Alibaba Cloud (MLX version by the MLX Community) · Apache 2.0. Translate and Enhance on your computer.
- VoxCPM2 by OpenBMB (MLX version by the MLX Community) · Apache 2.0. Closest-to-you English voice, and designing voices.
- Maya2 Native by Maya Research · Apache 2.0. Indian languages.
- Mimi audio codec by Kyutai · CC BY 4.0. The sound Maya2 speaks through. Used unchanged, under CC BY 4.0.
Tools inside the app
- FFmpeg by The FFmpeg developers · GPL 3.0. Reading and writing audio files. Runs as a separate program; source code at ffmpeg.org, and the licence text is inside the app (Contents/Resources/bin/FFMPEG-LICENSE.txt).
- MLX by Apple · MIT. Running models on Apple silicon.
- PyTorch by The PyTorch Foundation · BSD 3-Clause. Running the voice models.
Stentor also uses many open-source Python packages, each under its own licence, kept with the package inside the app. Questions about any of them: [email protected].