Demo of OVRLipSync
This project runs in Godot4.3 and uses a version of https://github.com/goatchurchprime/two-voip-godot-4 that is compiled against OVRLipSync, a library released by Oculus only on Windows and Android for generating visemes from audio WAV data.
FirstPastThePost Single viseme does a winner takes all filter where the viseme with the
highest score is set to 1 and all the others set to zero. This makes the animation jerky,
but you can see the mouth formations (particularly of the consonants) more clearly.
In the future we could ease between these shapes and sync it properly with some delayed audio.
Also, the laughter detection now makes an icon appear when it hits a threshold so you can test it by laughing at it.
On a platform without OVRLipSync you can tie the audio volume to a chosen viseme in the dropdown to animate the face.
There are bugs in the Godot Android AudioStreamMicrophone that intermittently turn the MicWorking off. Sometimes it makes an infinite loop of sound: godotengine/godot#95120
