Repolode
Submit
neuphonic

neuphonic/neutts

On-device open-source TTS speech models with instant voice cloning from about 3 seconds of audio, built on small LLM backbones.

Open-source alternative to:ElevenLabs

View on GitHub
Stars
6,322
Forks
661
Open issues
37
Last commit
3 months ago

What is it?

NeuTTS is a family of open-source text-to-speech models from Neuphonic that run on-device. Each model pairs a small LLM backbone with the NeuCodec neural audio codec, so speech is generated locally rather than through a web API. The models support instant voice cloning: a short reference clip and its transcript are enough to synthesize new text in that speaker's voice.

The lineup includes NeuTTS-Air for English, the NeuTTS-Nano collection with English, French, German and Spanish variants, and NeuTTS-2E, an English model that offers emotional control and four fixed speakers. Quantised GGUF builds are provided for running on phones, laptops and Raspberry Pis, and ONNX decoders are available for lower-latency inference. The repository includes Python examples for batch and streaming synthesis, reference encoding, and a finetuning script.

Generated audio includes a Perth watermark by default. The Python package is installed with pip, with optional extras for llama-cpp-python and onnxruntime.

Key features

  • 01Instant voice cloning from a short reference clip and its transcript.
  • 02GGUF quantisations for running on phones, laptops and Raspberry Pis.
  • 03Streaming synthesis with llama-cpp-python.
  • 04Multilingual Nano models for English, French, German and Spanish.
  • 05NeuTTS-2E offers emotion control across seven emotions and four speakers.
  • 06Watermarked output by default.

Use cases

  • Building embedded voice assistants or toys that run without a cloud service.
  • Adding offline speech output to apps where data must stay on the device.
  • Prototyping voice cloning with short reference recordings.
  • Running text-to-speech on low-power hardware such as Raspberry Pi.

Quick start

  1. 1

    Install NeuTTS from PyPI

    pip install neutts
  2. 2

    Clone the repository and enter the project folder

    git clone https://github.com/neuphonic/neutts.git && cd neutts
  3. 3

    Run the basic example to synthesize speech

    python -m examples.basic_example --input_text "My name is Andy. I'm 25 and I just moved to London. The underground is pretty confusing, but it gets me around in no time at all." --ref_audio samples/jo.wav --ref_text samples/jo.txt

Things to know

  • Some models use the NeuTTS Open License 1.0 rather than a standard open-source license, and the repository license is not clearly stated.
  • Streaming is only supported with GGUF models.
  • Non-English models work best with a reference clip in the same language.

In collections

AI video & audio tools

Topics

#text-to-speech#voice-cloning#on-device#speech-synthesis#gguf#llm#edge-ai

FAQ

What is neutts?
On-device open-source TTS speech models with instant voice cloning from about 3 seconds of audio, built on small LLM backbones. NeuTTS is a family of open-source text-to-speech models from Neuphonic that run on-device. Each model pairs a small LLM backbone with the NeuCodec neural audio codec, so speech is generated locally rather than through a web API. The models support instant voice cloning: a short reference clip and its transcript are enough to synthesize new text in that speaker's voice.
Is neutts free to use? What license does it have?
neutts is open source under the Other license. Check the license terms before using it in commercial products.
Is neutts still maintained?
Updated within 6 months. The last commit to neutts was on Jul 30, 2026.
What can I use neutts for?
Building embedded voice assistants or toys that run without a cloud service. Adding offline speech output to apps where data must stay on the device. Prototyping voice cloning with short reference recordings. Running text-to-speech on low-power hardware such as Raspberry Pi.
How do I install neutts?
Install NeuTTS from PyPI: pip install neutts Clone the repository and enter the project folder: git clone https://github.com/neuphonic/neutts.git && cd neutts Run the basic example to synthesize speech: python -m examples.basic_example --input_text "My name is Andy. I'm 25 and I just moved to London. The underground is pretty confusing, but it gets me around in no time at all." --ref_audio samples/jo.wav --ref_text samples/jo.txt
Where can I get neutts?
The source code and installation instructions are on GitHub: https://github.com/neuphonic/neutts

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects