Piano Fundamental Frequency Estimation Algorithm Based on Weighted Least Square Method

Fundamental frequency estimation(FFE) has been studied for many years, and current techniques are still not accurate and robust. Because of the fundamental frequency losing, the harmonic losing and the inharmonicity of piano sound, these FFE methods are invalid generally. The mean square error of the harmonic frequency based on quadratic interpolation of FFT is discussed, and the influence of other harmonic sidelobe on the harmonic measurement with the different length Hanning, Hamming, Chebyshev and Blackmanharris windows is analyzed. A FFE algorithm using weighted least square method is derived. The simulation experiment shows that the method is feasible, and the relative error is about 10^-5.