AuViMi stands for audio-visual mirror. The idea is to have CLIP generate its interpretation of what your webcam sees, combined with the words thare are spoken.
Why do you think that https://github.com/EvgenyKashin/random-colabs is a good alternative to AuViMi