Version: 1.18
PRELIMINARY/CONFIDENTIAL
TrueSpeech® Co-Processor
9.6
Setting the Base Frame Size
The CT8022 supports programmable base frame sizes to allow operation with different speech compression
algorithms that may operate on different fundamental frame sizes and frame periods (e.g. 30ms or 10ms). The
following table shows the base frame sizes recommended for different speech compression algorithms in
combinations with various other CT8022 features. Note that the higher processing overhead incurred when
operating on smaller speech frames leaves less processing resources available for the operation of additional features
like AEC. The fundamental frame and period sizes shown are the basic sizes of the corresponding speech
compression algorithm.
Speech
Compression
Type
Fundamental Fundamental Minimum Base Frame Minimum Frame Compressed
Algorithm
Frame Size
(Samples)
Frame Period Size Recommend For
Period
Recommended
For Operation
With The
Speech
Block Size
(16-Bit
Words)
(Milliseconds)
Operation With The
CT8022 (Samples)
See Note 1
CT8022
G.723.1 6.3/5.3
(all feature
240
240
240
80
30ms
30ms
30ms
10ms
240
240
240
80
30ms
30ms
30ms
10ms
12/10
16
combinations)
TrueSpeech 8.5
(all feature
combinations)
TrueSpeech
4.8/4.1 (all feature
combinations)
9/8
6
G.729A+B
8 KBPS
(includes 1
word frame
type indicator)
G.722 KHz
80
10ms
40
wideband 64KBPS
ADPCM
Note 1: For correct operation of the CT8022, the base frame size selected must be an integer multiple of the
fundamental frame size for the speech compression algorithm(s) used.
Note 2: Transfers of compressed speech data between the Host controller and the CT8022 should be made
via the CT8022’s Transmit/Receive Data Buffer. Use of the CT8015-style transfer protocol consumes
additional CT8022 processing resources during data transfer, leaving less processing power available
for compression + AEC etc.
G.729 Annex A operates at exactly 8.0 Kbits/sec. With a 10ms frame size, this rate should produce 80 bits every
10ms. 80 bits of data requires exactly five 16-bit words. However, with the inclusion of the G.729 Annex B
VAD/CNG silence compression, the combined G.729AB coder can produce three different types of speech frame.
These are the normal speech 80 bit speech frame, a 16-bit noisy silence frame, and a no transmit indication frame.
For the CT8022 to indicate the G729AB frame type, an extra frame type indicator word is added at the beginning of
each speech block so that the total size of each compressed speech block is 6 words instead of 5.
The base frame size selected affects the operation of the entire CT8022 device. The same base frame size is used for
both the record and playback channels.
The DTMF and Call Progress Tone detectors themselves are designed to operate with frame sizes of 30, 20 or 10ms.
The status from these detectors is updated and reported after every frame processed, regardless of the actual frame
size.
CT8022A11AQC FW Revision 0118 DSP GROUP, INC., 3120 SCOTT BOULEVARD
SANTA CLARA, CA 95054 PH: 408 986 – 4300 FAX: 408 986 – 4490
55
All specifications are subject to change without prior notice.
ETC [ ETC ]
