Real-Time Voice Conversion Developer

We're building a desktop app that converts a speaker's voice to a specific cloned voice in real time, with no noticeable lag, working inside Zoom, Teams, or Google Meet. A companion app runs on a laptop, captures the mic, converts speech to a fixed target voice locally on-device, and outputs it to a virtual microphone. The user selects that virtual mic in their call app, and the call proceeds normally — the other person hears the converted voice. What you'll build: Cross-platform app (Windows + macOS), real-time streaming voice conversion (RVC-style) Virtual audio device integration (VB-Cable/BlackHole) for Zoom/Teams/Meet Low-latency pipeline (10-40ms frames), crossfading to avoid artifacts Auto fallback if conversion lags; GPU/NPU acceleration a plus Simple installer/setup for non-technical users Target: under 200ms end-to-end added latency (ideally 90-170ms). Deliverables: working installers (Win/Mac), source + docs, latency benchmark report, user setup guide. Ideal skills: real-time audio/DSP, ML inference in production (ONNX/PyTorch), RVC/so-vits-svc experience, Tauri/Electron, virtual audio drivers, CoreML/DirectML a plus.