Audio
MOSS-TTS
MOSS-TTS family: open text-to-speech models for voice cloning, dialogue, voice design, and real-time agents
★ 4.2k starsApache-2.0
View guide →MOSS-TTS-Nano: a 0.1B open multilingual speech model that clones voices and streams in real time on CPU
MOSS-TTS-Nano is an open-source audio project by OpenMOSS (OpenMOSS/MOSS-TTS-Nano on GitHub). It has 4,457 GitHub stars and 557 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 6 September 2026.
Create the documented Python 3.12 Conda environment.
Clone the repo, install requirements, and install the package in editable mode.
Run inference with a reference audio clip and the text you want spoken; weights download automatically from Hugging Face.
Launch the web demo or CLI server for interactive use, or fine-tune on your own data with the included code.
A very small autoregressive speech model designed to run in real time even on CPU.
Multilingual synthesis with native stereo output at 48 kHz.
Clone a voice from a short reference recording passed as the prompt audio.
Streaming inference, including an ONNX runtime the project benchmarks at real-time on one CPU core.
Generate or serve from the `moss-tts-nano` CLI, a local web UI on port 18083, or Python scripts.
Training code is provided so you can adapt the model to your own voices or data.
Clone the project and enter its directory.
git clone https://github.com/OpenMOSS/MOSS-TTS-Nano.git
cd MOSS-TTS-NanoCreate the Conda env, install requirements, and install the package.
conda create -n moss-tts-nano python=3.12 -y
conda activate moss-tts-nano
pip install -r requirements.txt
pip install -e .Run inference with a reference clip (weights auto-download), or start the web demo with `python app.py`.
python infer.py --prompt-audio-path assets/audio/zh_1.wav --text "Hello from MOSS-TTS-Nano."OpenMOSS, the MOSI.AI research lab, which also publishes the larger MOSS-TTS model family.
Yes — code and weights are Apache-2.0 licensed and downloadable from Hugging Face.
No. It runs entirely locally; a GPU is optional since it is designed for CPU real-time use.
Nano is the 0.1B edge-friendly member of the family: far smaller and CPU-capable, trading some of the flagship 8B model's fidelity for speed and deployability.
Create the Python 3.12 Conda env, clone, `pip install -r requirements.txt && pip install -e .`, then run `python infer.py` with a prompt audio file — or `python app.py` for the web demo.
MOSS-TTS family: open text-to-speech models for voice cloning, dialogue, voice design, and real-time agents
Vietnamese TTS with instant voice cloning — on-device, real-time CPU inference, 48 kHz audio
Chatterbox — Resemble AI's open-source TTS family: expressive text-to-speech and voice cloning, multilingual V3, low-latency Turbo and tiny Nano models, MIT
A tiny conversational TTS model with cloning, streaming, and CPU speed.
On-device streaming speech-to-text engine for real-time apps.
Turn books and documents into natural speech with the Kokoro TTS model.
The most-starred other Audio projects in the LibreAI catalogue — the realistic shortlist if MOSS-TTS-Nano is not the right fit.
利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.
Robust Speech Recognition via Large-Scale Weak Supervision
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
Clone a voice from 5 seconds of audio and generate arbitrary speech in real time
Port of OpenAI's Whisper model in C/C++
Free and open — grab the source, star the repo, and start creating.