Voxveil
Real-time vocal reduction across desktop audio is technically demanding when the system must stay local, cross-platform, low-latency, and safe around OS audio routing.
An in-development local-first audio application exploring a shared Rust DSP core, platform-specific audio interception, and a responsive Tauri/React control surface.
In development
Voxveil is exploring real-time vocal reduction with a shared Rust audio core, native platform audio adapters, and a Tauri/React interface. The current implementation includes classic DSP center suppression, optional stem-agnostic AI interfaces, local-only state, and explicit routing rules.
On Windows, the project is working through multiple interception strategies—including an endpoint-scoped audio processing object and controlled virtual-endpoint relay paths—while guarding against stale routes, feedback loops, and unsafe sink selection.
This belongs in the Lab because the engineering surface is substantial and public, but there is no release yet. The portfolio will treat it as active work until a release establishes a completed public milestone.