Version: 1.18
PRELIMINARY/CONFIDENTIAL
TrueSpeech® Co-Processor
9.15.4
Adding Additional Echo Suppression
Additional echo attenuation can be achieved by increasing the level of echo suppression used by the CT8022. Echo
suppression is different from echo cancellation. In echo cancellation, only the echo component of the input signal is
affected, and any input signal that originates at the microphone input is not affected. With echo suppression the
entire input signal is affected, both the component due to echo and the component due to the real microphone input.
Echo suppression was the technique used in traditional half-duplex speakerphones. In this type of telephone, only
one audio signal direction is enabled at any given time. This takes advantage of the fact that in a normal
conversation only one user is talking at a given moment. This allows a decision to be made to attenuate one speech
direction, based on which speech direction is the inactive direction. The audio signal for the inactive speech
direction is always completely muted. To completely mute one direction of speech, half-duplex speakerphones
introduce greater than 60dB of attenuation into the inactive speech path.. In a half-duplex speakerphone, since only
one speech direction is enabled at any time, there is no possibility of hearing an echo. The problem with this
solution comes when both users attempt to talk at the same time. When this happens, the direction decision-making
logic in the half-duplex speakerphone is unable to function correctly and the result is often that both speech
directions are attenuated.
The CT8022 implements a scaleable form of Echo Suppression where the amount of attenuation added to the
inactive direction can be programmed in the range of 0 to -28dB. Using 0dB of echo suppression effectively
de-activates the echo suppressor. Providing scaleable echo suppressor attenuation allows a gradual compromise
between full-duplex operation (0dB suppression) and half-duplex operation (with -28 dB suppression). With a
scaleable echo suppressor, even when a speech direction is attenuated, it is not completely muted, so it is possible to
tell that speech is present.
By adding a small amount of echo suppression to the CT8022 (around 16 dB), it is possible to ensure that any
residual echo not fully cancelled by the AEC will be removed.
The echo suppression attenuation (loop attenuation) can be programmed only after the AEC is activated, and after
playback and record have been started. Use the Set Speakerphone command described in Section 9.21.3.
CT8022A11AQC FW Revision 0118 DSP GROUP, INC., 3120 SCOTT BOULEVARD
SANTA CLARA, CA 95054 PH: 408 986 – 4300 FAX: 408 986 – 4490
105
All specifications are subject to change without prior notice.
ETC [ ETC ]
