Experimental evaluation of echo path modeling with chaotic coded speech

摘要

An experimental evaluation of an echo-path modeling device is reported for chaotically-coded speech telephony circuits. The device provides an accurate characterization of the far-end echo path using only half-duplex near-end speech during conversational calls. The in-service nonintrusive systems of interest are usually based on a class of least-mean-square (LMS) digital adaptive filters (DAFs). The chaotic-based modulation regime exploits a logistic mapping to optimize the convergence rate of the model while preserving the signal bandwidth requirements. Experimental results show that uncoded speech is ineffective in driving the modeling device when noise-to-echo ratios exceed approximately -30 dB and thus invalidates live measurements. The chaotic modulation, however, increased the echo-to-noise ratio and allowed adaptation for far-end noise conditions exceeding 30 dBs higher than that of the uncoded case.