Realtime AI Voice Conversion & Cloning
A realtime AI voice conversion platform built on RVC. Converts live microphone input into a trained target voice in real time. Used to replicate Sir Jackie Stewart's voice for the Race Against Dementia charity campaign.
Creating custom voice content requires expensive voice talent or robotic-sounding TTS.
RVC Rox is a realtime AI voice conversion platform built on RVC (Retrieval-based Voice Conversion). Converts live microphone input into a trained target voice in real time. Proven in production — used to replicate Sir Jackie Stewart's voice for the Race Against Dementia charity campaign.
RVC WebUI (open source, GPU-accelerated) • RMVPE pitch extraction model for accurate realtime pitch tracking • HuBERT base model for speaker embeddings • PyTorch with CUDA 12.1 (NVIDIA GPU required) • VB-Audio Virtual Cable for routing converted audio to any application • Pipeline: mic input → pitch extract → RVC inference → virtual audio output
Everything you need to succeed
True realtime voice conversion at GPU speed
Train custom voice models from audio samples
Virtual audio routing — converted voice available in any app: Zoom, Teams, OBS, Synthflow
WebUI for model training, management and inference
Scriptable launch for live event or studio deployment
Pairs directly with Outreach Engine for custom-voice AI calling campaigns
Real-world applications
Campaign narration in a specific voice (e.g. Jackie Stewart for Race Against Dementia)
Brand voice consistency across AI-generated content
Custom voices for Synthflow AI calling agents
Live event and broadcast voice transformation
Personalised AI voice agents for client campaigns
Agencies, broadcasters, and content creators needing custom voice cloning and realtime conversion.
Open-source foundation with production-proven deployment for high-profile campaigns.
Choose the plan that fits your agency