What the player asked for, which is not always what they get: Auto takes Opus when the binaries are there and ADPCM when they are not, and an explicit Opus request on a build without them is refused rather than silently
FVoiceFragmentBuffer — short clips of what other players actually said, for a creature that answers in their voices. It keeps utterances, not a stream: a run of voiced frames becomes one fragment when the talker pauses,
FVoiceGrainFilter — the third layer of proximity voice, after distance and walls: the voice breaking up as it reaches the edge of what carries. Lower fidelity rather than lower level, so a far voice is still there but th
What the microphone path cleans up before a frame is encoded. Each stage can be switched off on its own: a headset user has no echo to cancel, a studio microphone wants no noise gate fighting it.
FVoiceRadioFilter — what makes a voice sound like it came out of a handheld: the telephone band, 300 Hz to 3.4 kHz. One low-pass cannot do it (the low end is the half that says "small speaker"), so this is a high-pass an
One speaker's decoding state. A stateful codec (Opus predicts each frame from the ones before it) needs a decoder per stream: a single decoder fed several speakers interleaved predicts every frame from someone else's voi
The player-facing voice controls for Visual Scripting and game UI: a scoreboard's mute button, a per-player volume slider, the mimicry opt-out. Until these existed a player had no way to reach SetPeerMuted at all. Every
VoiceChat/VoiceChatModule.h
Free functions
Signature
Description
void RegisterVoiceChatModule()
Registers the voice scene system (with its position/occlusion/visual-script-pin attach wiring) and the push-to-talk character extension. Idempotent. Called by the plugin loader's InitializeModule / the monolithic static registrar; headless tests call it directly.