Visit the Project Babble website ↗
A Windows app that bridges eye and face tracking to VRChat through compatible VRCFaceTracking SDK modules. Download modules for Babble Headset Runtime, Baballonia, Virtual Desktop, Steam Link, SRanipal, Bigscreen Beyond, or EyeTrackVR. Works with Jerry’s templates and VRChat’s native face-tracking SDK beta; available tracking depends on your hardware and module.
- Install the app and download a VRCFaceTracking module from Modules, or import a module DLL or ZIP.
- In Modules, choose Eyes & eyebrows and Face & mouth, then click Apply sources. Start each source’s headset runtime or desktop tracking app.
- Enable OSC in VRChat and select a face-tracking avatar.
Use the same module for both, or combine eyes from one module with the face from another. Selected badges show each module’s eye/face assignment. A shared module runs once; separate modules feed one combined VRChat output. Use module assigns that module to both roles. Your source pair and each module’s settings are remembered.
Downloads come from the module authors; modules are never bundled. The bridge can only send expressions your sources measure. Known overlapping input ports are rejected; give the modules and their tracking apps different input ports before combining them. Remove modules with the trash button. Removing either selected source stops the bridge and clears both roles; your other modules stay installed.
Using Baballonia Expressions Alpha? Download Baballonia in Modules, or choose Update for Expressions Alpha on an older imported module.
In Settings, choose whether to start with SteamVR, close when it exits, or start minimized to system tray. Clicking X keeps tracking running in the tray; double-click the tray icon to reopen, or right-click it to Quit.
Click the kitty face in the bottom-left corner to meet Relay, an optional holographic robot cat. Relay starts off every time you open the app. Gently stroke it with your pointer to pet it, click to poke it, or drag its body to pick it up. It also climbs visible panels and perches on their edges. A local million-parameter Transformer guides its behavior using recent context without blocking the interface. A small personal learning layer adapts to gentle interaction, and it can discover and practice varied hops, rollovers and hopping runs. Its individual temperament, growing bond, preferences and shared memories are saved on your computer and loaded next time. In Settings, Forget memories starts that bond fresh. Relay is completely silent: cat sounds appear only as little text bubbles. The companion pauses in the tray. Behavior, memory and research notes.
The tracking-connection particles smoothly coast to a stop and dim when the bridge stops, then brighten and resume when it starts.
Other compatible VRCFaceTracking SDK modules work too. module.json is optional; use a ZIP if a module needs supporting files.
Only import modules from publishers you trust: they run code on your PC under their own licenses.
Windows 10/11 x64. Setup & requirements · What’s new · Build from source
Original code: Apache 2.0. Copyrights & third-party licenses.
Pull requests welcome · Report a security issue privately
Independent project; not endorsed by Project Babble, VRCFaceTracking, Valve, or VRChat.