3D printed body. Tried various airflow sensors including small dynamic microphones and sensitive piezo film sensors. Decided on 40Khz tiny disc transducers in each of 12 chambers, 1 transducer in front, one at back of chamber. Measure time of flight for 40Khz bursts, did not need to be too accurate to sense airflow. PIC32 based, Sense blown airflow ultrasonically per chamber . Simple s/w synth, PWM out, low power audio amp, small speaker. Worked OK, synth needs more work & tuning. Spit covering disc transducers when playing was an unexpected issue, was resolved by moving transducer out of the way to top of chambers and angled to avoid 'spit splatter'. The microphones and piezo film sensors were badly affected by spit. Actually spend more time on the airflow sensors than the synth s/w eventually. Conclussion was that moisture resistant air pressure sensors will work better due to also being able to pick up airflow presence and strength per chamber.