State of the Art in Acoustic Echo Cancellation

Abstract

Hands-free speech communication, i.e. speaking and listening through an audio terminal device without the help of a hand-set, is a concept which appeared many years ago. Indeed, the problem of acoustic echo control is crucial for the design of audio terminals operating in hands-free mode such as hands-free telephone sets, teleconference systems, videoconferencing devices…. The acoustic echo control can be performed by more or less elaborate devices. The simplest implementations make use of echo suppressors which introduce variable losses on the send and receive paths: these devices do not allow a real full-duplex communication and often fail to operate in noisy environments. A clearly better solution consists in using an adaptive echo canceller, which constructs a replica of the actual acoustic echo path by means of an adaptive filtering method, in a way similar to “electric” echo cancellers installed in the telephone network to get rid of the hybrid mismatches [86]. Despite its apparent simplicity, acoustic echo control is presumably one of the most difficult and challenging signal processing tasks. Many research efforts are presently ongoing on this topic in communication and signal processing laboratories around the world. More than ninety publications mainly from the last two and a half years support this statement [34, 35].