查看文章

Much interest has recently been shown in sequential adaptation methods for APC systems. Such methods offer a trade-off between speed of convergence and misadjustment. Inadequacies in the former become most apparent when the speech waveform changes abruptly following certain consonants or unvoiced utterances. This paper offers a method of speeding up convergence at these points by causing the adaptive filter to operate on samples one block ahead of those to be transmitted. Additional time is thus given for adaptation at these critical points.