Improved Humming Music Retrieval Method Based on Wavelet Transformation and Dynamic Time Warping

This paper proposes a method, which makes use of the band-pass property of the wavelet transform, combining the wavelet analysis with autocorrelation function to extract melody characters from the humming signal. After the wavelet transform, the humming signal is weighed, and then combined with auto-correlation function for signal analysis. The weighted factors in the design, gives full consideration about that the sound of men and women with different frequency range. This paper adopts the improved DTW algorithm to match melodies, and presents one kind of music melody representation irrespective with tone. While carrying through the DTW algorithm, it can avoid the up and down transition of tone and also reduce the calculation of melody matching.