Open-source credits
Thank you to the people who build and share these.
Software
- OpenVoice v2 (MyShell, MIT) — "my own voice" tone conversion on your device
- Kokoro-82M (hexgrad, Apache-2.0) — natural backup voice on your device for English, Spanish, French, Italian and Portuguese
- Piper (Rhasspy / Open Home Foundation, MIT) — backup voice on your device; loader from vits-web (Diffusion Studio, MIT)
- eSpeak NG (GPL-3.0) — pronunciation for the backup voice, loaded unmodified from @diffusionstudio/piper-wasm; source: github.com/espeak-ng/espeak-ng
- Depth Anything V2 Small (Apache-2.0) — camera depth motion on your device
- FLUX.2 [klein] (Black Forest Labs, via Cloudflare Workers AI) — consistent characters across scenes
- ONNX Runtime Web (Microsoft, MIT), MediaPipe (Google, Apache-2.0), mp4-muxer (MIT)
Backup voices (Piper)
- Turkish "fahrettin" — dataset CC0 (Nabu Casa voice-datasets)
- English "ljspeech" — LJ Speech dataset, public domain
- German "thorsten" — Thorsten Müller, CC0
- Spanish "davefx" — CC0
- French "siwis" — SIWIS French Speech Synthesis Database, Junichi Yamagishi et al., CC BY 4.0
- Italian "paola" — Paola Persico, CC0
- Portuguese "faber" — CC0
- Russian "denis" — CC0
- Ukrainian "ukrainian_tts" — CC0
Media
Stock photos and clips: Pixabay Content License, Pexels License. Emoji: Noto Emoji (Google, Apache-2.0).