Hume provides human ratings and APIs for evaluating voice, speech, and conversational AI.
RVC V2
RVC WebUI is a framework for training voice conversion models with short voice recordings.
About RVC V2
RVC V2 is an open-source Python framework for training voice timbre conversion (voice changer) models using short voice recordings. It provides a web-based interface for model training and inference, plus a separate real-time voice changing GUI with a reported default end-to-end latency of 170ms (90ms when using ASIO input/output devices, dependent on hardware driver support). The framework uses a pre-trained base model built from nearly 50 hours of high-quality VCTK dataset audio, and includes licensed song datasets for training to reduce copyright risk for users. It supports CPU, AMD/Intel, and NVIDIA GPU hardware, with tailored dependency packages for different configurations, and can train functional models with as little as 10 minutes of low-noise speech.
Key features
- Reduces tone leakage via top1 retrieval of training set features
- Trains models with as little as 10 minutes of low-noise speech
- Supports model fusion to blend timbres via checkpoint merging
- Includes built-in pymss/MSST model for vocal/instrument separation
- Uses InterSpeech2023-RMVPE for high-pitch voice extraction
- Provides WebUI for training/inference and real-time voice changing GUI
Best for: It is best suited for users seeking to train custom voice timbre conversion models with limited voice data, including those using lower-end or non-NVIDIA hardware.
Good for
- Supports model training with as little as 10 minutes of low-noise speech
- Compatible with CPU, AMD/Intel, and a range of NVIDIA GPUs, including lower-end hardware
- Includes pre-trained base models and licensed song datasets to reduce training time and copyright risk
Watch out for
- Real-time voice changing default latency is 170ms, with 90ms only available via ASIO devices dependent on hardware driver support
- The upcoming RVCv3 pre-trained base model is not yet released for use
- AMD/Intel Linux systems are limited to CPU acceleration with no GPU acceleration option
RVC V2 at a glance
- Category
- AI & Machine Learning › AI Voice & Audio
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Yes · Source code
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
RVC V2 alternatives
Tools and sites you could use instead, ranked by how closely they match.
Create AI voice assistants for business phone calls with Awaz’s no-code voice platform.
Free trial
24 alternatives
MMAudio synthesizes audio from video and text through a multimodal joint-training approach presented at CVPR 2025.
Open source
19 alternatives
The KittenTTS project offers a text-to-speech model described as under 25 MB in size.
Open source
24 alternatives
EmotiVoice provides a multi-voice text-to-speech engine with prompt controls.
Open source
24 alternatives
Applio provides free open source voice conversion, custom voice model training, and real time conversion.
Open source
24 alternativesPopular in AI Voice & Audio
Well-known listings from the same topic.
Google Illuminateilluminate.google.comAlteredaltered.aiELSA Speakelsaspeak.comKreadoAIkreadoai.comPodwisepodwise.aiAddsubtitleaddsubtitle.ai
RVC V2 FAQ
How much voice data is required to train a voice conversion model?
A minimum of 10 minutes of low-noise speech is recommended for training.
What hardware configurations are supported for RVC V2?
It supports CPU, AMD/Intel systems (Windows uses DirectML, Linux uses CPU) and NVIDIA GPUs, with separate dependency packages for different hardware.
What is the latency of the real-time voice changing feature?
The default end-to-end latency is 170ms, reducible to 90ms when using ASIO input and output devices.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.