Musical pitch tracking using internal model control based frequency cancellation

A new method for pitch estimation of musical sound signal is presented in this paper. This method is based on the behavior of a notch filter in an error feedback system, and was first developed for identification of periodic signals with uncertain frequency. Unlike many previous methods, which are based on frequency representation in time window of the music signal, our algorithm is based on instantaneous ´measurements´ of the frequency in real-time. Because of the high accuracy of the frequencies and the magnitudes we obtained, this method may also be used to verify the types of the instruments and even the vibratos. Simulation results show that the presented approach operates reliably in monophonic case with a highly accurate frequency estimation. The same method has been extended to the polyphonic setting. At present, we are restricting input with a relatively strong fundamental frequency. This paper is a progress report, we hope to extend our work greatly in the future.