Version: 1.18
PRELIMINARY/CONFIDENTIAL
TrueSpeech® Co-Processor
To disable AS/AB:
C1
S1
C2
S2
E013H
E013H
0000H
0000H
The NLP/AS/AB commands may be sent at any time, before or after the Acoustic Echo Canceller is activated. It is
necessary to send these commands only once. It is not necessary to send them every time the AEC is started, but,
sending them more than once will not cause any problems.
9.15.5.3
Dynamic Activation of NLP and Attenuation Scaling (AS)
The NLP and AS/AB features are designed to be activated only in the presence of far-end speech or echo. The
CT8022 offers two mechanisms for activating the NLP/AS/AB features:
•
•
NLP/AS/AB activation controlled by speech direction
NLP/AS/AB activation when the measured AEC cancellation exceeds the attenuation lock threshold
Refer to Section 9.15.6.3 for details concerning the cancellation threshold used for activating NLP/AS/AB.
Refer to Section 9.15.6 for details concerning the NLP/AS/AB activation mode
9.15.5.4
Sample Advanced AEC Configurations
The CT8022 offers flexibility in controlling the advanced AEC features (NLP/AS/AB) to allow customization of
AEC performance to meet the requirements of different echo environments.
9.15.5.4.1
Acoustic Echo Environments
Acoustic Echo environments tend to show the following typical properties:
•
•
•
•
•
High levels of background noise
Very short direct (microphone-speaker) echoes with short durations (< 5ms) with high amplitude echoes
Significantly extended multi-path echo durations (> 50-100ms) producing very low amplitude echoes
Increased non-linearity in the echo path
Low input user signal levels relative to speaker output levels (user is far from microphone, but speaker and
microphone are close together)
•
Echo path varies in real-time requiring rapid AEC training
9.15.5.4.2
Telephone Hybrid Echoes
Telephone Hybrid (2-to-4 wire interface) echo environments tend to show the following typical properties:
•
•
•
•
•
Low noise levels
Short echo durations (< 10ms)
Very little non-linear distortion in the echo path
Input (microphone) and Output (speaker) signal levels are approximately equal
Echo path is constant for a single call
CT8022A11AQC FW Revision 0118 DSP GROUP, INC., 3120 SCOTT BOULEVARD
SANTA CLARA, CA 95054 PH: 408 986 – 4300 FAX: 408 986 – 4490
109
All specifications are subject to change without prior notice.
ETC [ ETC ]
