neuphonic/neutts

On-device TTS model by Neuphonic

What it solves

NeuTTS is designed to move high-quality text-to-speech (TTS) from cloud-based APIs to local devices. It provides a set of lightweight, natural-sounding speech models that can run on phones, laptops, and embedded devices like Raspberry Pis without requiring a constant internet connection, ensuring better privacy and lower latency.

How it works

The system uses a combination of a small LLM backbone (optimized for text understanding) and a neural audio codec called NeuCodec. This architecture allows the model to process text and generate audio tokens that the codec then converts into audible speech. It supports instant voice cloning, where a short (3-15 second) audio sample and its corresponding text are used as a reference to mimic a specific voice.

Who it’s for

It is built for developers creating embedded voice agents, local AI assistants, smart toys, and compliance-safe applications where on-device processing is critical.

Highlights

  • On-Device Optimization: Provides GGUF quantizations for efficient inference on mobile and embedded hardware.
  • Instant Voice Cloning: Ability to clone a voice using as little as 3 seconds of audio.
  • Multilingual Support: Models available for English, Spanish, German, and French.
  • Emotional Control: The NeuTTS-2E model allows for specific emotional expressions (e.g., happy, sad, angry) for fixed speakers.
  • Real-time Performance: Capable of real-time generation on mid-range devices.
  • Built-in Security: Includes perceptual threshold watermarking on generated audio to prevent misuse.

Related

  • Project
    sipeter/CloneTTS

    An offline Android TTS engine that enables local voice cloning from short audio samples for private, on-device text-to-speech synthesis.

  • Project
    index-tts/index-tts

    A zero-shot text-to-speech system that clones voices from a single audio clip with fine-grained control over emotion, speed, and pronunciation across five languages.

  • Project
    resemble-ai/chatterbox

    A family of open-source text-to-speech models providing low-latency voice cloning and multilingual speech generation across various model sizes for GPU and CPU deployment.

  • Project
    Audio8-AI/Audio8_TTS

    A compact 0.6B-parameter multilingual text-to-speech model featuring zero-shot voice cloning and efficient deployment options for CPU and GPU.

  • Project
    chenpipi0807/ComfyUI-Index-TTS

    A ComfyUI extension that implements IndexTTS for high-quality voice cloning and text-to-speech, featuring multi-lingual support, emotion control, and multi-character novel parsing.