You can overcome the RPi scarcity by migrating the code to the Teensy platform, which aside being cheaper and less power hungry than the 2,3,4 RPi, is a lot cheaper and more easily available. Not an easy task since there's no Linux under the hood, but there are some excellent audio/midi libraries to help. They already built commercial-level synthesizers with it. By combining the breath sensor data with other pressure sensors you could end up with a very expressive instrument.
https://www.pjrc.com/teensy/index.html
https://www.pjrc.com/teensy/td_libs_Audio.html