Sorry, I'm not qualified to speak about this. You may find an answer from the maintainer email address listed on ALDEx2's bioconductor page.
I don't know enough R to help with this. If you haven't already I would suggest walking through the GLM example in the ALDEx2 reference manual (see the top of page 13) and observing what the input variables used (selex, covariates, and mm) all look like. This should be a good starting point for seeing how your data will eventually need to be formatted. You may also be interested in the awesome qiime2R package, which was designed for loading QIIME 2 outputs like your ASV table in R.
From looking over section 5.6 of the vignette, I think you're on the right track -- covariates seems to be analogous to an ordinary sample metadata file (loaded as an R data.frame), and mm is a model matrix (aka a design matrix) that's created based on an R formula and covariates.