Music signal Separation based on improved Multipitch estimation method using Hidden Markov Model

Multiple Pitch Estimation (MPE) of polyphonic music is one of the most important and difficult issues in the area of the Music Information Retrieval (MIR). The main job of MPE is to estimate both the accurate value and the number of the fundamental frequencies. In this paper, we first discuss harmonic reset algorithm and octave correction algorithm. Harmonic reset algorithm can separate overlapping harmonics of multiple notes and octave correction algorithm can correct Sub frequency and frequency doubling errors. Combined with fundamental frequency candidates, harmonic reset and octave correction modules and preprocessing and post-processing modules, this paper presents a method of multi-fundamental frequency estimation for single frame signal. Using harmonic product spectrum to solve candidate fundamental frequency sets, the computational complexity is greatly reduced. Harmonic reset and octave correction significantly improve the accuracy, thus achieving a balance between low computational complexity and high accuracy. A multi-base frequency estimation method for multi - frame signals is proposed based on a chord identifier and a hidden Markov model.