LaziBook — model files

Model files used by the LaziBook app (offline e-book reader and audiobook player for iOS, macOS and Android). Each folder keeps the licence of its original model.

app/ — bundled in the app (ios/Models, Android voice_* packs)

Folder	Model	Source	Licence
app/turbo/	VieNeu-TTS v3 Turbo (backbone MLP layers 1–10 4-bit, the rest 8-bit)	pnnbao-ump/VieNeu-TTS-v3-Turbo	Apache-2.0
app/nano/	VieNeu-TTS v3 Nano: text encoder, codec decoder (int8), voices. The estimator and duration predictor are LaziBook's own trained weights and are kept in a private repo	pnnbao-ump/VieNeu-TTS-v3-Nano	Apache-2.0
app/clone/	Nano encoders for voice cloning (codec, reference, speaker, denoiser)	VieNeu-TTS v3 Nano	Apache-2.0
app/codec/	MOSS Audio Tokenizer Nano (ONNX)	MOSS-Audio-Tokenizer-Nano-ONNX	Apache-2.0
app/g2p/	sea-g2p (Vietnamese G2P)	pnnbao97/sea-g2p	Apache-2.0
app/phonon/	Phonon 31M English, Core ML 4-bit palettized, precompiled .mlmodelc + 38 voices	gradium/phonon-31m-en, gradium/phonon-voices	Apache-2.0
app/vitszh/	VITS Chinese (sherpa-onnx vits-zh-ll)	k2-fsa/sherpa-onnx TTS models	see source
app/matcha/	Matcha-TTS zh-en + Vocos 16 kHz (sherpa-onnx)	dengcunqin/matcha_tts_zh_en_20251010	see source

Optional downloads

Folder	Used by	Source	Licence
ming-3.0-tts-0.5B-4bit/	Mac app, Chinese/English voice pack	jorgepb96/Ming-3.0-tts-0.5B-4bit @ d06229c, from inclusionAI/Ming-3.0-tts-0.5B	MIT
mt/	Offline translation (Hy-MT2 1.8B 1.25-bit GGUF)	tencent/Hy-MT2-1.8B-1.25Bit-GGUF @ 9df5c82	Apache-2.0