Analysis of quantization errors in subband speech coding with modified DFT filter banks

Abstract

The modified DFT filter bank is a highly efficient pseudo-QMF filter bank with almost perfect reconstruction, which is introduced in [N.J. Fliege, Modified DFT polyphase SBC filter banks with almost perfect reconstruction, in: Proceedings of the ICASSP, Australia, April 1994, vol. 3, pp. 149–152] and discussed in [N.J. Fliege, Computational efficiency of modified DFT-polyphase filter banks, in: Proceedings of the 27th Asilomar Conference on Signals, Systems and Computers, Asilomar, USA, November 1993; T. Karp, N.J. Fliege, MDFT filter bank with perfect reconstruction, in: Proceedings of the IEEE International Symposium on Circuits and System, Seattle, USA, May 1995]. Because of its comparable computational efficiency with cosine modulated filter banks and almost perfect reconstruction at the critical sampling rate, the MDFT filter bank can be used in subband speech coding. This paper aims at a theoretical analyses of the quantization errors in the subband speech coding with MDFT filter banks, and discusses how to design the prototype filter for an MDFT filter bank in order to minimize the quantization errors. A simplified and suboptimal procedure for the design of the prototype filter is presented.