Title :
Architecture of a single chip acoustic echo and noise canceller using cross spectral estimation
Author :
Liem, Murco ; Manck, O.
Author_Institution :
Inst. of Comput. Eng. & Microelectron., Berlin Univ. of Technol., Germany
Abstract :
This paper describes the architecture of a single chip acoustic echo canceller and noise filter for full duplex hands free communications. To our knowledge the CENS90200 is the first ASIC to eliminate the acoustic echo by using cross spectral estimation. This approach is much more robust against the environmental noise typically found in a hands free (e.g. automotive) environment than the usually employed class of "normalized least mean square" (NLMS) algorithms. Additionally a noise filter reduces the background noise by using spectral minimal statistics and diffusive gain factors. These advanced algorithms have substantially greater computational performance requirements than their traditional counterparts. A highly portable architecture is presented that meets these requirements at low clock to sample frequency ratios. It combines directly hardware mapped algorithms and a purpose built processor optimized for spectral signal processing, thus making a low cost, single chip implementation possible.
Keywords :
acoustic signal processing; application specific integrated circuits; digital filters; digital signal processing chips; echo suppression; spectral analysis; speech processing; ASIC; CENS90200; acoustic echo elimination; automotive environment; clock to sample frequency ratios; computational performance requirements; cross spectral estimation; diffusive gain factors; environmental noise; full duplex hands free communications; hardware mapped algorithms; noise filter; portable architecture; purpose built processor; single chip acoustic echo canceller; spectral minimal statistics; spectral signal processing; Acoustic noise; Application specific integrated circuits; Automotive engineering; Background noise; Echo cancellers; Filters; Noise cancellation; Noise robustness; Signal processing algorithms; Working environment noise;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). 2003 IEEE International Conference on
Print_ISBN :
0-7803-7663-3
DOI :
10.1109/ICASSP.2003.1202447