Turning Your Wind Instrument into a Music Controller: Real-time Fingering Estimation by Classifying Reflected White Noise

We propose a novel fingering estimation method that allows a player to use any wind instrument as a music controller by attaching a single microphone and loudspeaker-embedded mouthpiece. The loudspeaker plays white noise continuously while the fingerings are estimated in real-time based on the sound pressure recorded at the end of the instrument. Our method addresses a major problem of conventional music controllers: differences in the tactile feel of keys compared to the player’s own instrument. We demonstrated that the proposed method accurately estimated fingerings on a saxophone with promising performance (a 1.05 % misclassification rate), satisfying the low-latency feedback required in the context of musical performance.