The HITar is still at the prototype stage. As a project within the Artificial Intelligence and Music programme at Queen Mary University of London, we actively conduct research to further refine its workings with iterative development and user studies. Currently, the AI algorithm runs on a laptop, but it was designed and coded so that migration to embedded hardware will be fast.
Work is under way to support and explore even richer ways to represent hand hits: by using more than two hand parts, therefore supporting more than two MIDI notes, and by involving localisation (an indication of where the guitar body was hit) in the description parameters.
At its heart, our technology is quite versatile: it is a method for low-dimensional description of impact sounds with very low latency. We intend to explore its application to many other percussion techniques beyond the gestures that our models were trained on, as well as its use with sticks and mallets rather than hands. Its uses beyond percussion, or indeed beyond musical applications, should also be explored too.