I haven't tried your app, but from the video it seems you have those discrete notes. Have you though about having a sort of continuous time/frequency domain? I guess you could do that with an Inverse Wavelet transform or Inverse Short Time FT with logarithmic frequency scaling. I'm assuming your sounds are synthesized, not recorded.

Another idea - how about making a part of the app sort of "real time" (or short delay), like an instrument? Seems it would take the intuitiveness and freshness of gesturing to a whole new level.