Skip to content
Get packs

GPT-SoVITS

Open-source tool that clones a voice from about a minute of audio and speaks typed text in it.

GitHub Download 20250606v2pro
Stars
64.6k
Forks
6.8k
Licence
MIT
Language
Python
Last update
5 days ago
Type
App

What it is

GPT-SoVITS is a Python project with a browser-based WebUI for text-to-speech and voice cloning. You can give it a short voice sample for instant (zero-shot) speech, or fine-tune a model on about one minute of recorded data for a closer match to the speaker. The README also describes built-in helpers for separating vocals from accompaniment, splitting audio into training segments, Chinese speech recognition, and labeling text.

  • Generate speech from text using a short reference clip of a voice
  • Fine-tune a model on about one minute of training audio for a closer voice match
  • Separate vocals from background music in recordings using the built-in tools
  • Create a training dataset by automatically segmenting audio and labeling text

Great for

  • Hobbyists who want to experiment with cloning a voice they have permission to use
  • Creators who need narration in a consistent synthetic voice
  • Developers who want a self-hosted TTS system with a WebUI
  • Users comfortable following Python environment setup steps

Not ideal for

  • People who need a zero-install tool, since setup requires Conda, Python, and a capable GPU for best results
  • Anyone wanting to clone a voice without the speaker's consent, which the README does not address beyond licensing
  • Users needing guaranteed real-time speed on modest hardware, since the README's speed figures come from high-end GPUs
Step-by-step guide: how to use GPT-SoVITSInstall, first run, examples, tips and FAQs
GPT-SoVITS alternativesBark vs GPT-SoVITS

How to get it

  1. 1

    Clone and build from source

    You'll need: Conda · Python 3.10 (3.10 to 3.12 supported per README) · FFmpeg · A CUDA-capable GPU is recommended; Apple silicon and CPU are also listed

    Clone the repository

    git clone https://github.com/RVC-Boss/GPT-SoVITS

    Create and activate a Conda environment (Linux)

    conda create -n GPTSoVits python=3.10

    Activate the environment

    conda activate GPTSoVits

    Run the Linux installer

    bash install.sh --device <CU126|CU128|ROCM|CPU> --source <HF|HF-Mirror|ModelScope> [--download-uvr5]
  2. 2

    Run it

    • Install prerequisites and run the installer for your platform
    • Download the pretrained models as described in the README, unless install.sh completed them
    • Start the WebUI using the launch method described in the README
    • Upload a short voice sample and enter text to generate speech
  3. 3

    Get the most out of it

    • Use the Docker images and Docker Compose setup if you prefer a containerized environment
    • Set is_half to true on GPUs that support half precision to reduce memory use
    • Use the built-in vocal separation and dataset segmentation tools to prepare cleaner training data
    • Try the Lite Docker variants if you do not need the full feature set

Downloads

Source code (.zip)main branch · straight from GitHubRelease 20250606v2proRelease notes and source

Strengths

  • Supports few-shot training from roughly one minute of data according to the README
  • Includes a WebUI so beginners do not need to write code to use it
  • MIT licensed with a large, active user base

Trade-offs

  • Setup involves Conda, Python 3.10 to 3.12, FFmpeg, and typically a GPU, which is demanding for non-experts
  • The README notes that Mac GPU-trained models are lower quality, so CPU is used there instead
  • Output quality and speed depend heavily on hardware and training data; the README's figures are from specific test machines

How it compares

ProjectStarsTypeLevelUpdatedBest for
GPT-SoVITS (this)64.6kAppintermediate5 days agoHobbyists who want to experiment with cloning a voice they have permission to use
bark39.3kAI model / agentintermediate2.2 years agoDevelopers who want to add text-to-speech or sound generation to a Python project
OpenVoice38.1kAI model / agentadvanced1.5 years agoDevelopers who want to build voice cloning or text-to-speech features into their own projects
piper11.3kAppintermediate1.2 years agoDevelopers who want offline speech output in a local app or home automation setup
voicebox59.2kAppbeginner30 days agoPeople who want text-to-speech or voice cloning without sending audio to cloud services
Retrieval-based-Voice-Conversion-WebUI39kAppintermediate3 months agoHobbyists who want to make cover songs or voice-changed clips with a trained voice

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.