Repolode
Submit
resemble-ai

resemble-ai/chatterbox

Open-source text-to-speech model family by Resemble AI, with voice cloning from a short reference clip across English and many other languages.

Open-source alternative to:ElevenLabs

View on GitHubWebsite
Stars
27,100
Forks
3,649
Open issues
375
Last commit
3 months ago

What is it?

Chatterbox is a family of open-source text-to-speech models from Resemble AI. The models generate speech from text and can clone a voice from a reference clip supplied at generation time. The family includes Chatterbox-Turbo, a 350M-parameter English model aimed at low-latency voice agents, Chatterbox-Nano, a 110M-parameter model that targets CPU and on-device inference, and Chatterbox Multilingual V3, a 500M-parameter model covering 23+ languages. A Single Language Pack of dedicated finetunes for specific languages and regional dialects is also available.

Turbo and Nano support paralinguistic tags such as [laugh], [cough] and [chuckle] in the input text to add non-verbal sounds. The multilingual model takes a language ID with each request, and the original Chatterbox model exposes CFG and exaggeration parameters for tuning delivery. Python installation is available through pip, and the project includes example scripts for TTS and voice conversion. Every generated audio file carries an imperceptible PerTh neural watermark, and a watermark extraction script is included.

Key features

  • 01Voice cloning from a short reference audio clip
  • 02Turbo and Nano models with paralinguistic tags like [laugh] and [cough]
  • 03Multilingual model covering 23+ languages with language ID selection
  • 04Nano model runs on CPU for on-device use
  • 05Built-in PerTh watermarking on generated audio
  • 06Installable via pip with Python examples

Use cases

  • Building low-latency English voice agents for phone or customer support calls
  • Generating narration and creative voiceovers with expressive non-verbal sounds
  • Localizing voice content across multiple languages with consistent voice identity
  • Running speech synthesis on CPU or edge devices with limited memory

Quick start

  1. 1

    Install the package from PyPI

    pip install chatterbox-tts
  2. 2

    Load a model and generate speech from a text prompt with a reference clip (see the Usage examples in the README)

Things to know

  • The README states it was developed and tested only on Python 3.11 on Debian 11.
  • Turbo and Nano are English-only, and multilingual use requires the separate multilingual models.

Topics

#text-to-speech#voice-cloning#speech-synthesis#multilingual#python#pytorch#audio#voice-agent

FAQ

What is chatterbox?
Open-source text-to-speech model family by Resemble AI, with voice cloning from a short reference clip across English and many other languages. Chatterbox is a family of open-source text-to-speech models from Resemble AI. The models generate speech from text and can clone a voice from a reference clip supplied at generation time. The family includes Chatterbox-Turbo, a 350M-parameter English model aimed at low-latency voice agents, Chatterbox-Nano, a 110M-parameter model that targets CPU and on-device inference, and Chatterbox Multilingual V3, a 500M-parameter model covering 23+ languages. A Single Language Pack of dedicated finetunes for specific languages and regional dialects is also available.
Is chatterbox free to use? What license does it have?
chatterbox is open source under the MIT license. Check the license terms before using it in commercial products.
Is chatterbox still maintained?
Updated within 6 months. The last commit to chatterbox was on Jul 21, 2026.
What can I use chatterbox for?
Building low-latency English voice agents for phone or customer support calls Generating narration and creative voiceovers with expressive non-verbal sounds Localizing voice content across multiple languages with consistent voice identity Running speech synthesis on CPU or edge devices with limited memory
How do I install chatterbox?
Install the package from PyPI: pip install chatterbox-tts Load a model and generate speech from a text prompt with a reference clip (see the Usage examples in the README)
Where can I get chatterbox?
The source code and installation instructions are on GitHub: https://github.com/resemble-ai/chatterbox

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

Related projects