In a hands-free mode of speech communication, a loudspeaker produces
acoustic energy of a distant signal from a far-end talker and a
microphone produces a near-end signal containing a speech component
representing the speech activity of a near-end talker or an acoustic echo
component, or both. An echo replica is produced from the distant signal
and a residual echo representing the difference between the near-end
signal and the echo replica. The residual echo is used as a feedback
signal to produce the echo replica. Using one of the near-end signal and
the residual echo as a first input signal and the echo replica as a
second signal, an estimate of the acoustic echo is produced when the
speech activity of the near-end talker is low or zero. Using the acoustic
echo estimate, the spectrum of the first input signal is shaped to
produce a local signal for transmission to the far-end talker.