Voice Chat

Normcore includes high-quality, low-latency voice chat that's accessible via a single component: RealtimeAvatarVoice.

RealtimeAvatarVoice works on any realtime prefab as long as the RealtimeView component has an owner.

Upon instantiation, RealtimeAvatarVoice uses the ownership of the RealtimeView to determine if it should be sending audio for this prefab or receiving audio to play back.

If the component is owned locally, a microphone stream is created, and if it is owned by a remote client, an AudioOutput component is created, in order to play back audio.

By default RealtimeAvatarVoice will create an AudioSource component, however, if you would like to set custom spatialization settings, create an AudioSource manually. RealtimeAvatarVoice will detect and use this instance instead.

RealtimeAvatarVoice includes a mute property. If RealtimeAvatarVoice represents a local player, this property will stop sending audio, effectively muting this player for everyone in the room. However if RealtimeAvatarVoice represents a remote player, it will mute their audio stream only for the local client.

RealtimeAvatarVoice also includes a voiceVolume property. This is a read-only value between 0-1 that reflects that audio level the last time it was sampled for playback. This can be useful for animating the mouth of an avatar to match a player's speech.

AudioPreprocessor

RealtimeAvatarVoice includes an AudioPreprocessor that automatically sets the gain for the microphone and filters out background noise, reverb, and echo cancellation.

In most cases the default settings are correct, but if you would like to make changes, the source code to RealtimeAvatarVoice is included and can be forked.

It's worth noting that most platforms default to using hardware echo cancellation supplied by the operation system, so the AudioPreprocessor's echo cancellation functionality is not required.