Intriguing concept and application of AI here. While the sound itself is quite basic, I see potential in this being explored further for accessibility, compositional, or research reasons. Beware that it can soft lock when playing the highest or lowest notes. Nice work!