Limits on echo return loss enhancement on a voice coded speech signal

Abstract

Acoustic echo cancellation is a desirable feature for small cellular handsets that have significant acoustic coupling from speaker to microphone. Current echo cancellation techniques prefer an echo canceller to be placed at the wireless handset. Since echo cancellation requires much processing power it is beneficial to move the echo canceller to the base station, where power consumption will not be a constraint. This paper examines the feasibility of performing linear echo cancellation on a signal coming from a mobile at the base station. The mobile signal at the base station contains speech coding which may degrade the effectiveness of echo cancellation. The performance of linear echo cancellation on mobile echo encoded and decoded by low bit rate speech coders operating at 4 kbits/sec to 13 kbits/sec, including GSM full rate, G.723a, G.729, and AMR speech coders, was investigated. The best echo return loss enhancement achieved depends on the bit rate of the coder, and is usually less than 10 dB for the low bit rate operating range examined.