CosyVoice

Free plan APILinuxSelf-hosted

What CosyVoice is

CosyVoice is a multilingual text-to-speech system for inference, training, and deployment, available as free, open-source software. Fun-CosyVoice 3.0 is designed for zero-shot multilingual speech synthesis across nine languages and more than 18 Chinese dialects or accents, with multilingual and cross-lingual zero-shot voice cloning. Users can guide language, dialect, emotion, speed, and volume, and adjust pronunciation with Chinese Pinyin or English CMU phonemes. It handles numbers, special symbols, and varied text formats without a traditional frontend module. Text-in, audio-out streaming has stated latency as low as 150 ms, and WAV is a listed export format. The repository documents installation with Conda and Python 3.10, model downloads through ModelScope or Hugging Face, and Docker deployment using gRPC or FastAPI. It also documents NVIDIA Triton and TensorRT-LLM acceleration. The repository is Apache License 2.0 licensed and describes its content as intended for academic purposes and demonstrating technical capabilities. Installation and deployment are self-managed, and no paid plans are listed.

Who it suits

CosyVoice suits developers or technical teams seeking a self-managed multilingual speech system with training and deployment options. Its stated academic purpose may also fit users exploring speech-system capabilities.

What it does well

Free and Apache License 2.0 licensed.
Supports voice cloning and nine languages.
Provides text-in, audio-out streaming.
Offers FastAPI and gRPC deployment interfaces.
Supports WAV output.

What to know first

Installation uses Python 3.10.
Installation and deployment are self-managed.
Content is described as intended for academic purposes and demonstrating capabilities.

Plans

Plan	Billed and limits	Price
Open-source software	The repository provides downloadable source code and models; no paid plans are listed. · Apache-2.0 licensed repository · self-managed installation and deployment	Free

As github.com lists them, read 3 Oct 2026. The maker's own page has the final word.

Getting started

Open the CosyVoice GitHub repository.
Install using the documented Conda setup and Python 3.10.
Download pretrained models through ModelScope or Hugging Face.
Choose a deployment route such as Docker with gRPC or FastAPI.
Use the documented FastAPI or gRPC server and client interfaces.

Verdict

CosyVoice offers multilingual speech synthesis, voice cloning, and several documented deployment paths as free software. Its self-managed setup and stated intended use are worth considering before adopting it.

Questions

Does CosyVoice have a free plan?

Yes. The repository lists open-source software at no cost, with downloadable source code and models and no paid plans listed.

What license does CosyVoice use?

The repository is licensed under Apache License 2.0.

Which platforms are listed?

The listed platforms are API, Linux and self-hosted.

Can CosyVoice be used commercially?

Commercial use is listed as yes.

What installation setup is documented?

The repository documents Conda installation with Python 3.10. An optional ttsfrd normalization wheel is specified for Linux x86_64.

Where can users get support?

The maintainers direct users to GitHub Issues and an official Dingding chat group.

Facts from CosyVoice's own pages

API access: Yes · github.com, 20 Sept 2026
API interfaces: The deployment instructions include FastAPI and gRPC servers and clients. · github.com, 3 Oct 2026
Commercial use: Yes · github.com, 20 Sept 2026
Deployment options: The repository documents Docker deployment with gRPC or FastAPI, as well as NVIDIA Triton and TensorRT-LLM acceleration. · github.com, 3 Oct 2026
Export formats: WAV · github.com, 20 Sept 2026
Free plan: Yes · github.com, 20 Sept 2026
Installation: The repository documents installation with Conda and Python 3.10, and offers pretrained model downloads through ModelScope or Hugging Face. · github.com, 3 Oct 2026
Intended use: The repository says its content is for academic purposes and to demonstrate technical capabilities. · github.com, 3 Oct 2026
Languages and dialects: Version 3.0 covers nine languages and more than 18 Chinese dialects or accents, with multilingual and cross-lingual zero-shot voice cloning. · github.com, 3 Oct 2026
License: The repository is licensed under Apache License 2.0. · github.com, 3 Oct 2026
Platforms: linux,api,self_hosted · github.com, 20 Sept 2026
Pronunciation control: The system supports pronunciation inpainting with Chinese Pinyin and English CMU phonemes. · github.com, 3 Oct 2026
Purpose: CosyVoice is a multilingual text-to-speech system that provides inference, training, and deployment capabilities. · github.com, 3 Oct 2026
Runtime requirements: The documented installation uses Python 3.10; the optional ttsfrd normalization wheel is specified for Linux x86_64. · github.com, 3 Oct 2026
Streaming: It supports text-in and audio-out streaming, with latency stated as low as 150 ms. · github.com, 3 Oct 2026
Support: The maintainers direct users to GitHub Issues and an official Dingding chat group for discussion. · github.com, 3 Oct 2026
Text normalization: It supports reading numbers, special symbols, and varied text formats without a traditional frontend module. · github.com, 3 Oct 2026
vLLM compatibility: The repository says CosyVoice 2 and 3 support vLLM 0.11.x or newer and vLLM 0.9.0, while versions between those releases are untested. · github.com, 3 Oct 2026
Voice cloning: Yes · github.com, 20 Sept 2026
Voice instructions: Users can provide instructions for language, dialect, emotion, speed, and volume. · github.com, 3 Oct 2026
Zero-shot synthesis: Fun-CosyVoice 3.0 is designed for zero-shot multilingual speech synthesis. · github.com, 3 Oct 2026