Skip to content
RepoQuarry

GPT-SoVITS RVC-Boss/GPT-SoVITS

Few-shot voice cloning and TTS WebUI; 5-second zero-shot or 1-minute fine-tune, five languages, MIT code and weights.

GPT-SoVITS at a glance

Checked 15 Oct 2026

WebUI for zero-shot TTS from a 5-second sample or fine-tuning on 1 minute of audio, with cross-lingual output in English, Japanese, Korean, Cantonese and Chinese. Bundles dataset tools (vocal separation, slicing, ASR for labelling). MIT code and MIT pretrained weights on Hugging Face. Runs on NVIDIA, AMD (ROCm on Linux), Apple silicon and CPU; stated RTF 0.028 on an RTX 4060 Ti and 0.526 on an M4 CPU. Windows has a download-and-double-click package.

Cite this pageRepoQuarry. “GPT-SoVITS.” Checked 15 Oct 2026. https://repoquarry.dev/projects/gpt-sovits

How much of ElevenLabs does it replace?

2 of 4 jobs. Strong cloning and TTS, but only five languages (EN, JA, KO, Cantonese, ZH) versus a broad multilingual catalogue.

Partial: Speech generation Partial replacement
Partial: Voice cloning Partial replacement
No: Speech recognition Not covered
No: Voice agents Not covered

For speech recognition, voice agents, see other ElevenLabs alternatives. You may also need to build an interface, an API server and hosting yourself.

How much of Descript does it replace?

2 of 3 jobs. Voice cloning and text-to-speech only; no transcript-based audio or video editing.

No: Transcript-based editing Not covered
Partial: Voice cloning Partial replacement
Partial: Speech generation Partial replacement

For transcript-based editing, see other Descript alternatives. You may also need to build an interface, an API server and hosting yourself.

Facts, with sources

Every answer links to where we found it. We read the documentation; we did not install it.

Report a correction
QuestionAnswerSourceChecked
Runs on my machine? Confirmed: Yes, local WebUI on GPU, Apple silicon or CPU; Docker images available If you are a Windows user (tested with win>=10), you can download the integrated package and double-click on go-webui.bat to start GPT-SoVITS-WebUI. github.com · GPT-SoVITS#tested-environments ↗ 1d ago
Code licence Confirmed: MIT MIT License. Copyright (c) 2024 RVC-Boss github.com · LICENSE ↗ 1d ago
Model weights licence Confirmed: MIT (lj1995/GPT-SoVITS pretrained models); optional extras such as UVR5 and ASR models are separate downloads not checked here license: mit huggingface.co · GPT-SoVITS ↗ 1d ago
Commercial use Confirmed: Yes, MIT code and core weights; check licences of optional third-party models you add (UVR5, Damo ASR, Faster Whisper) pretrained models used in https://github.com/RVC-Boss/GPT-SoVITS huggingface.co · GPT-SoVITS ↗ 1d ago
Paid services needed? Confirmed: None required Download pretrained models from GPT-SoVITS Models and place them in GPT_SoVITS/pretrained_models. github.com · GPT-SoVITS#pretrained-models ↗ 1d ago
Hardware Partial: No minimum RAM/VRAM stated. Stated speed (v2 ProPlus): RTF 0.028 on RTX 4060 Ti, 0.014 on RTX 4090, 0.526 on an M4 CPU. Tested on CUDA, Apple silicon and CPU. 0.028 tested in 4060Ti, 0.014 tested in 4090 (1400words~=4min, inference time is 3.36s), 0.526 in M4 CPU. github.com · GPT-SoVITS#readme ↗ 1d ago
Operating systems Confirmed: Windows (prebuilt package or install.ps1), Linux (CUDA, ROCm, CPU), macOS (MPS or CPU) Note: The models trained with GPUs on Macs result in significantly lower quality compared to those trained on other devices, so we are temporarily using CPUs instead. github.com · GPT-SoVITS#installation ↗ 1d ago
Setup effort Confirmed: Easy on Windows (integrated package); conda plus install script on Linux and macOS If you are a Windows user (tested with win>=10), you can download the integrated package and double-click on go-webui.bat to start GPT-SoVITS-WebUI. github.com · GPT-SoVITS#windows ↗ 1d ago

Get the weekly top 5 and what’s new, every Monday

We’ll email you once a week. Unsubscribe any time. We never share your email.

Cookie preferences

Choose what we may store in your browser. You can change this any time from Cookie settings in the footer.

Essential

Security, and remembering this choice. rq-consent · 12 months

Always on