The therapy involves getting patients to shout in a certain way and try and reach certain volume levels or follow certain volume patterns. E.g. shout gently for five seconds and gradually ramp up the volume over 10 seconds.
It would be really easy to replicate. I reckon the most time consuming part would be testing and calibration.
The whole app directory is less than 2MB.
But you should look into Max/MSP ( https://cycling74.com/products/max/ ), where you could build and bundle a simple audio app like that in minutes and have it tailored to your specific needs.