I am a Electronics and Communication Under Grad Engineer. Please suggest me some project ideas related with Nvidia graphics processors, CUDA and Deep learning using Nvidia processors.
Implementing a MIMO (Multiple Input/Multiple Output) transmission chain in the acoustic range could be fun. Set up 4 speakers and 4 mics, point the speakers in different directions (and possibly add some cardboard for a much more directional response of the mics).
Then see if you can exploit this spatial diversity with some good processing at both the sender and the receiver end.