TTS with mouseless calibration

[theaspect]

This is additional demo page demonstrating Augmentative and Alternative Communication
Initial calibration and choosing color with text-to-speech ability without user input
Red dot now moving with ease-in-out animation and not teleporting to next position
All zones of interest working as magnet to ease choosing and re-calibrating on successful fixation
I have more ideas to implement

I'll rebase after #164 merge

[xanderkoo]

Hi @theaspect, thanks for your help. I've just merged #164 and #185, so you can go ahead and rebase and I'll review the PR.

[theaspect]

@xanderkoo I rebased, but see serious face detection degradation even on site examples: face mesh continuously wiggling and warming up to green state takes much more time

Also something has changed with aspect ratio, need to figure out what.

[xanderkoo]

@theaspect Thanks! I'm not seeing the degradation -- is that with the demos on your repo? I wonder if it has to do with a local installation of TensorFlow or something, since we're using Facemesh face detection out of the box.

The green box in the video preview is just meant to verify that the user's eyes are within the central region of the camera, which could indeed be affected if your Facemesh isn't working properly.