NeuTTS
A collection of open-source, on-device text-to-speech models with instant voice cloning from a few seconds of audio, distributed in GGUF format for phones, laptops, and Raspberry Pis.
NeuTTS is a collection of open-source, on-device TTS speech language models with instant voice cloning, built from small LLM backbones paired with NeuCodec, a 50Hz neural audio codec that reaches high audio quality at low bitrates using a single codebook. It aims to bring natural-sounding, real-time speech synthesis to a local device instead of a web API.
The lineup includes NeuTTS-Air for English, a multilingual NeuTTS-Nano collection covering English, French, German, and Spanish, and NeuTTS-2E, which adds emotional control. Quantized GGUF versions are available for on-device deployment on phones, laptops, or Raspberry Pis, a new speaker can be cloned from as little as three seconds of audio, its context window covers about thirty seconds of audio, and its outputs are watermarked.