Love seeing consult with rather than speak for!
To add to not detract from the important initial target group, I can hear, but I use captions for fatigue-related cognitive/processing needs.
Sorry I can't help much on resources etc. I do a tiny amount for work using provided tools that basically automate then I correct for accuracy, but that's almost exclusively pure speech.
I think context matters for describing sounds, just like for alt text. Transcribe the info the sound adds.