GPT-SoVITS

Free plan APILinuxMacSelf-hostedWebWindows

What GPT-SoVITS is

GPT-SoVITS is a WebUI for voice conversion and text-to-speech, with tools for preparing training data and an API for inference. It supports zero-shot speech generation from a five-second vocal sample; fine-tuning with one minute of training data can improve voice similarity and realism. Cross-lingual inference covers English, Japanese, Korean, Cantonese, and Chinese. The WebUI includes accompaniment separation, automatic training-set segmentation, multilingual speech recognition, and text labeling. Recognition options include Fun-ASR-Nano, SenseVoice, classic FunASR, and Faster Whisper. Its API provides GET and POST inference endpoints that return WAV audio streams on success. The project provides installation instructions for Windows, Linux, and macOS, plus Docker Compose services for CUDA 12.6 and 12.8 in full and Lite variants. The Lite image excludes ASR and UVR5 models; UVR5 models require manual download, while ASR models download as needed. The software is free, permits commercial use, and is released under the MIT License. Its guides are available in Chinese and English.

Who it suits

It may suit people creating voice models or speech from short voice samples, including beginners using the dataset tools. It is available to users who can install it on Windows, Linux, or macOS, or deploy it with Docker.

What it does well

Free plan and commercial use are allowed.
Five-second samples support zero-shot text-to-speech.
One minute of data can be used for fine-tuning.
Supports five languages for cross-lingual inference.
Includes GET and POST inference API endpoints.

What to know first

Lite Docker omits ASR and UVR5 models.
UVR5 models require manual downloading in Lite.
macOS temporarily uses CPUs due to reported GPU-model quality limits.
No SECURITY.md policy or published advisories are reported.

Getting started

Visit the GPT-SoVITS GitHub repository.
Choose installation guidance for Windows, Linux, or macOS, or use a documented Docker Compose service.
On Windows 10 or newer, download the integrated package and start the WebUI with go-webui.bat.
For the Lite Docker image, download UVR5 models manually if needed; ASR models download as needed.
Use a five second vocal sample for zero-shot text-to-speech, or prepare one minute of training data for fine-tuning.
Consult the Chinese or English user guide.

Verdict

GPT-SoVITS combines short-sample speech generation with fine-tuning, dataset tools, and an inference API. Check the Lite image omissions and macOS limitation against your setup before choosing a deployment.

Questions

Is GPT-SoVITS free?

Yes. It has a free plan and is released under the MIT License.

Can it be used commercially?

Yes. Commercial use is permitted.

Which languages does cross-lingual inference support?

English, Japanese, Korean, Cantonese, and Chinese.

Which operating systems are supported?

Installation instructions cover Windows, Linux, and macOS.

What does the Lite Docker image include?

It does not include ASR or UVR5 models. UVR5 models must be downloaded manually, while ASR models download as needed.

What audio formats are listed for export?

WAV, OGG, and AAC are listed.

Facts from GPT-SoVITS's own pages

API: The repository includes an API exposing GET and POST inference endpoints that return WAV audio streams on success. · github.com, 1 Oct 2026
API access: Yes · github.com, 20 Sept 2026
ASR integrations: The WebUI lists Fun-ASR-Nano, SenseVoice, and classic FunASR; Faster Whisper is also available as an ASR backend. · github.com, 2 Oct 2026
Commercial use: Yes · github.com, 20 Sept 2026
Dataset tools: WebUI tools include accompaniment separation, automatic training-set segmentation, multilingual ASR, and text labeling. · github.com, 2 Oct 2026
Deployment: The project documents Docker images and Docker Compose services, including full and Lite variants. · github.com, 2 Oct 2026
Docker: Docker Compose defines full and Lite services for CUDA 12.6 and CUDA 12.8 environments. · github.com, 1 Oct 2026
Export formats: wav,ogg,aac · github.com, 20 Sept 2026
Few-shot TTS: The project says users can fine-tune with one minute of training data to improve voice similarity and realism. · github.com, 2 Oct 2026
Free plan: Yes · github.com, 20 Sept 2026
Intended users: The integrated WebUI tools are described as assisting beginners in creating training datasets and GPT/SoVITS models. · github.com, 1 Oct 2026
Languages: Cross-lingual inference supports English, Japanese, Korean, Cantonese, and Chinese. · github.com, 2 Oct 2026
License: The repository identifies its license as MIT. · github.com, 2 Oct 2026
Lite limit: The Lite Docker image does not include ASR or UVR5 models; UVR5 models must be downloaded manually and ASR models download as needed. · github.com, 2 Oct 2026
Local API: The repository includes api.py and api_v2.py alongside the WebUI. · github.com, 2 Oct 2026
macOS limit: The README says models trained with Mac GPUs produce significantly lower quality than models trained on other devices, so it temporarily uses CPUs instead. · github.com, 2 Oct 2026
macOS limitation: The project says models trained with GPUs on Macs have significantly lower quality and therefore temporarily uses CPUs on macOS. · github.com, 1 Oct 2026
Operating systems: Installation instructions are provided for Windows, Linux, and macOS. · github.com, 2 Oct 2026
Platforms: web,windows,macos,linux,api · github.com, 20 Sept 2026
Product: GPT-SoVITS is a WebUI for few-shot voice conversion and text-to-speech. · github.com, 2 Oct 2026
Security policy: GitHub reports that the project has no SECURITY.md security policy and no published security advisories. · github.com, 1 Oct 2026
Support material: The README links to Chinese and English user guides. · github.com, 1 Oct 2026
User guide: The README links to Chinese and English user guides. · github.com, 2 Oct 2026
Voice cloning: Yes · github.com, 20 Sept 2026