Does this generalize to basically anything outside of speakers though? I think a pretty easy solution is to not use AI for that specifically.