- Kokoro-82M (voice model for text to speech): Apache-2.0
- kokoro-js (voice runtime): Apache-2.0
- phonemizer (text to sounds, for the voice): Apache-2.0
- eSpeak NG (compiled into phonemizer): GPL-3.0-or-later
- The voice runtime /vendor/kokoro/kokoro.js as a whole (kokoro-js, phonemizer with eSpeak NG, Transformers.js and ONNX Runtime), with the exact sources and build steps: GPL-3.0-or-later
- Whisper base (speech recognition model): MIT
- Transformers.js (runtime for the voice and speech recognition models): Apache-2.0
- ONNX Runtime Web: MIT
- Llama 3.2 1B (AI writing model). Built with Llama: Llama 3.2 Community License
- Qwen3.5 2B (the default AI writing model): Apache-2.0
- Qwen3 4B (AI writing model): Apache-2.0
- WebLLM (runtime for the AI writing models): Apache-2.0
- Vazirmatn (interface font for Persian, Sorani Kurdish and Urdu): SIL Open Font License 1.1
- IBM Plex Sans Arabic (interface font for Arabic): SIL Open Font License 1.1
- Heebo (interface font for Hebrew): SIL Open Font License 1.1
- Noto Sans Devanagari (interface font for Hindi and Marathi): SIL Open Font License 1.1
- Noto Sans Bengali (interface font for Bengali): SIL Open Font License 1.1
- Noto Sans Thai (interface font for Thai): SIL Open Font License 1.1
- Inter (font for our share images and screenshots): SIL Open Font License 1.1
- THIRD-PARTY-NOTICES.txt
Los modelos de escritura con IA no forman parte del código de este sitio. Este sitio los sirve sin modificar, desde los mismos archivos que las demás herramientas de Scissor: Llama 3.2 1B con la Llama 3.2 Community License (Built with Llama), y Qwen 3.5 2B y Qwen 3 4B con la Apache License 2.0. Cada uno se usa con su propia licencia, como explican nuestras condiciones de uso.