A Music Transcription Algorithm Applying State Estimation and Parameter Identification on the Time-frequency Plane
Yukio Fukayama, Daisuke Tanaka
An algorithm is proposed for a music transcription system that displays notes and rests on a music staff. It is able to process multivoice music, which includes chords and counterpoint. The system first projects the sound onto the time-frequency plane by applying the Gabor wavelet transform, and then identifies tempo and duration by autocorrelation of the projection. Next, it uses the fundamental frequencies of simultaneously played tones to estimate the note. A state estimation technique is used to handle instrument-dependent harmonics.