#6060th of 67 in VoiceFor voice, speech and music
neutts
On-device text-to-speech models with voice cloning from short reference audio
- Stars
- 6.3k
- License
- custom license
- Last commit
- Jul 2026
- Language
- Python
5,049 on 4 Apr 20266,338 on 31 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
NeuTTS is a family of small LLM-backbone speech models (about 120m to 360m active parameters) that synthesize speech and clone a voice from a 3-15 second reference clip. Models ship as PyTorch weights and Q4/Q8 GGUF files, paired with the NeuCodec audio codec, and run through a Python package. Variants cover English, French, German and Spanish, plus NeuTTS-2E with six emotions plus neutral.
Who it is for: Developers embedding local TTS with voice cloning in apps or devices
Strengths
- GGUF Q4/Q8 builds run on CPU; README benchmarks include a Galaxy A25 phone and an M4 iMac
- Voice cloning from about 3 seconds of reference audio, with pre-encoded sample voices bundled
- Streaming via llama-cpp-python, ONNX codec decoder for lower latency, and a finetune script
- Generated audio carries a Perth watermark by default
Weaknesses
- Context window is 2048 tokens, roughly 30 seconds of audio including the prompt
- Nano and 2E models use NeuTTS Open License 1.0; only Air is Apache 2.0
- Streaming works only with GGUF backbones, which need llama-cpp-python built for your hardware
- NeuTTS-2E has four fixed speakers and no cloning; uv sync install skips watermarking
What it needs
- GPU optional
- Needs llama-cpp-python, onnxruntime, PortAudio, OpenBLAS
- Models: NeuTTS-Air, NeuTTS-Nano, NeuTTS-Nano French/German/Spanish, NeuTTS-2E, GGUF Q4/Q8
Quick answers
Is neutts open source?
Its code is public on GitHub under its own license terms; read them before you use it at work.
Does neutts need a GPU?
No. It runs on a CPU and can use a GPU when you have one.
Also in Voice
See all 67| Rank | Project | Score |
|---|---|---|
| 1 | voicestudioDesktop app for voice cloning, dubbing, dictation and audiobooks | 97 out of 100 |
| 2 | whisper.cppC/C++ port of OpenAI Whisper for local speech-to-text transcription | 89 out of 100 |
| 3 | handyOffline push-to-talk dictation app that pastes transcripts into any text field | 89 out of 100 |
| #4 | xiaozhi-esp32ESP32 firmware for a voice chatbot with MCP device and cloud tool control | 82 out of 100 |
| #5 | buzzOffline desktop app for transcribing and translating audio with Whisper | 81 out of 100 |