Most robot faces only animate when they talk. People judge whether they are being heard while they are the ones talking, and that is where a frozen face fails. A few rules from character animation and conversation cover most of it:
- Hold the gaze, with small saccades. The eyes stay on the person but shift slightly every half second or so.
- Tilt a little. A small head tilt on entering the listening state reads as attention.
- Blink less. Attentive listeners blink less often; thinking raises the rate again.
- Nod at the pauses. When the person stops after speaking for a while, a small nod, and sometimes a blink, says "go on".
- Look away to think. While the reply is being worked out, the eyes go up and to one side. Staring straight ahead while thinking reads as frozen.
- Come back before speaking. A quick dart away and back, then the first word.
The engine does all of this from three calls: setState("listening"), setAudioLevel(micLevel, "mic") each frame, and setState("thinking") while your model writes its reply.
Get the engine, £29 Compare it with a mood swap