Comment by rdrdg

Currently, we are consider image, video and audio data types. This multi-modal idea sounds interesting none the less, thanks for sharing :)