This paper describes the use of two techniques to reduce the complexity of a Multi-Channel Echo Canceller (MCEC) system. First, the MCEC system complexity can be reduced by efficient utilization of DSP architectural features to achieve an increased number channels per DSP. The second technique relies on the fact that the adaptation procedures for the filter coefficients are suspended based on the speech activity factor. We present a method to estimate the reduction in complexity for a typical MCEC system based on the above model. Small performance degradation of fixed complexity systems dependent on speech statistics is analyzed and estimated. Finally, we submit results for performance of such systems based on simulations.
展开▼