@jscholes Actually my cards are mostly audio based as well because I've been using comprehensible input and there was a recommendation to avoid reading in the target language until fairly late in the process.
The cards are pretty Android specific and they currently only read the Spanish part, but they might be a good starting point. DM me your info and I can try to send you some of my stuff.
@pganssle Do you have an automated workflow for taking the LLM output and building it into cards? And if so, is it published anywhere?
I'm approaching the stage of learning where this sort of thing will be helpful, although I do better with audio-based cards at the moment as a blind screen reader user. I've just picked up an electronic braille display though to incorporate more touch reading.