#audio-only are the ones where the media is only audio, no graphics. #ASMR, for example.
#audio-based and #action-based describes if the script strokes to the audio (music) or actions in the video, usually used alongside #pmv and #hmv.
#audio and #action I believe are abridged version of the above 2 tags, often used incorrectly (e.g. ppl may tag #audio-only content with #audio, “action-packed” scenes with #action).
I once made these two synonyms of #audio-based and #action-based. Now I think they should just be deleted to avoid confusion - but I don’t know how to.
We also had a tag wiki back then, it may be outdated but some still make sense.