CT8022 Datasheet, PDF (107/194 Pages) List of Unclassifed Manufacturers – VOIP/VON G.723.1, G279AB TRUESPEECH CO-PROCESSOR

Version: 1.18
PRELIMINARY/CONFIDENTIAL
TrueSpeech® Co-Processor
9.15.5 Activating the Advanced Echo Canceller Features in the CT8022
Output
Signal
Echo
Signal
Echo
Estimate
Speech
Direction
Sensor
+^-
Attenuation
Scaling
NLP
AEC
Cancellation
Measurement
Echo
Suppressor
Figure 9-4: Advanced AEC
To increase the amount of echo attenuation provided by the AEC, the Host may enable the Advanced AEC features
of the CT8022. The Advanced AEC is actually comprised of two independent mechanisms that can be used to
reduce the residual echo left after actual cancellation.
• Non-Linear Processing (NLP)
• Attenuation Scaling or Boosting (AS/AB)
The NLP and AS/AB features are available both in DSVD and Standalone Speakerphone Mode.
9.15.5.1 Non-Linear Processing (NLP)
NLP introduces deliberate distortion into the signal so that low-level signals are distorted to a greater extent than
high-level signals. The CT8022 offers two NLP modes:
• center-clipping
• center-subtraction
Center-clipping uses the following algorithm:
if input is < +Threshold and input > -Threshold, then input is forced to zero.
Center-subtraction uses the following algorithm:
if input is > +Threshold then input = input - Threshold
else if input < -Threshold the input = input + Threshold
else input is forced to zero
Center-clipping produces a quantization distortion effect for small amplitude signals. Center-subtraction produces an
effect similar to cross-over distortion in a push-pull amplifier.
CT8022A11AQC FW Revision 0118
DSP GROUP, INC., 3120 SCOTT BOULEVARD
107
SANTA CLARA, CA 95054 PH: 408 986 – 4300 FAX: 408 986 – 4490
All specifications are subject to change without prior notice.