It does sound really cool. I just don't get why the shorthand part is important. Why not jot "grandma" and "poodle" and then tap the pen there for when the interviewee starts talking about his grandma and his poodle.
I'd actually really like something similar with a computer while I type and record.
It wouldn't be hard to throw something together, either. Really I just need it to pull a timestamp from the recorded audio after every word I type and associate some meta data and the timestamp with that word location.
Might be a fun after-work project. I'm sure there are already solutions like that out there, though