As people age or are exposed to loud sounds, high frequency hearing loss occurs. Most people can't hear above about 20kHz so almost nobody can hear bats. A solution to both problems is frequency shifting or compression. Both are easily done with fast Fourier transforms (FFT) with shifting optimally being done by octaves to preserve harmonic structure and moving, for instance, 2500 to 5000 to 1250 to 2500, 5000 to 10000 to the same range, and 10000 to 20000 to ditto, etc. This assumes that the user has "normal"hearing to 2500 Hz (in my case, I have pretty normal hearing to about 3500, so the range might be adjustable but preserve the harmonic structure).
Compression is similar with, perhaps, a difference in ranges -- lets assume we compress 1000 to 20000 to 1000 to 2500? Again, this should be configurable.
The problem on the Pi is audio input/output as this is not natively supported very well. Some additional hardware is needed. The FFT can probably be done just in software but there are APIs for using the video processing chip to speed up FFT. This is probably overkill but there don't seem to be any practical examples of using this API. The new Pi should be a great platform as it can run on battery for (apparently) a long time. Portability is a huge plus!
Stereo sound could be achieved by mounting a couple of small mics on a headset.
