Filtering optimization in MDCT domain base on spectrum character of HRTF

Abstract

An efficient filtering algorithm of multichannel virtual surround processing in modified discrete cosine transform (MDCT) was proposed.The MDCT filter bank can be represented in a polyphase structure and further derived to a high order matrix representation.By using the property of sparse matrix representation and the large dynamic range of head-related transfer function (HRTF),an adaptive neighborhood algorithm was presented for fast filtering with HRTF in the MDCT domain.The width of a neighborhood can be adjusted according to the spectrum of HRTF in MDCT domain.Experimental results show that the proposed algorithm greatly reduced the implementation cost,and maintained same filtering quality.In order to compare with different multichannel virtual surround processing methods,a subjective listening test was conducted.The listening results reveal that the proposed approach can achieve basically same subjective quality as the traditional filtering methods.The complexity analysis of the virtual surround processing methods indicates that the proposed filtering algorithm of HRTF consumes minimal calculation,which is particularly suitable for audio compressed in MDCT format.