So this is one of those good news, bad news situations. When life gives you lemons, design phased array acoustic systems!
Attendance was light at the first office hours roundtable, so I ended up developing some theory from an idea I had the night before. The premise is that if you have an array of microphones on the ceiling of a loud venue, you can do channelization, triangulation and signal filtering to identify the different speakers and the different conversation topics.
I think the next step in this project is to start working on simulation in something like Blender, where a number of conversations can be recorded with different speakers and different content, and then they can be played out with some simulated room noise (audacity?), which allows for testing of the signal processing chain. If you're interested and want to work together, please let me know!
