期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2011
卷号：11
期号：4
页码：85-90
出版社：International Journal of Computer Science and Network Security
摘要：Many researchers investigated sound source separation as one of themes of music information processing. Sound source separation is a work to separate each sound from a mix of several sound sources such as pianos and violins etc. It is one of very important techniques in automatic music transcription. And also it is assumed as one of early stages in music recognition of human auditory perception and it will take a major role in computer simulations of human perception process. We propose a new method of sound source separation using Butterworth parallel band pass filter(BPF). We show that the proposed method is superior to a recent work. As a result of experiment the proposed method improves fidelity of output comparing an original single instrument sound in input.
关键词：sound source separation; FFT; RCF; Butterworth filter; automatic music transcription
