Version: 1.18
PRELIMINARY/CONFIDENTIAL
TrueSpeech® Co-Processor
9.7.6
Enable VAD
The G.723.1/TrueSpeech 6.3, 5.3, 4.8 and 4.1 rates support Voice Activity Detection (VAD) and Comfort Noise
Generation (CNG) as defined in the ITU-T G.723.1 Annex A specification. This provides a form of built-in silence
compression. On the speech encode (record) side, the VAD feature can be selectively enabled or disabled. On the
decode (playback) side, CNG is automatic and is triggered by specially formatted speech frames that are generated
by the encoder when VAD is enabled. By default, the encoder VAD is disabled.
Note: The size of the physical data transfer performed for each speech frame when reading or writing to
the CT8022 is determined by the initial speech mode selection. For example, if the record session is
started in 6.3 mode, then all data transfers will consist of 12 word frames. The VAD/CNG frames
containing a smaller amount of significant data (2 words or 2 bits) have the unused portions of the
frame padded with zeros.
To enable VAD:
Command
Status
513FH
513FH
To disable VAD:
Command
Status
513EH
513EH
9.7.7
Dynamic Switching between TrueSpeech/G.723.1 6.3 and 5.3
The CT8022 supports on-the-fly switching between TrueSpeech 6.3 and 5.3 data rates as required by the ITU
G.723.1 specification. To enable switching between these two rates, the RECORD operation must be started with
the 6.3 rate selected. This sets the frame size used for CT8022-to-Host transfer to 12 words. When dynamically
switching between the 6.3 and 5.3 rates the frame size remains fixed throughout the record process. The 12-word
frame size is used for both the 6.3 and 5.3 speech data frames, although the 5.3 speech frame only contains 10 words.
This prevents synchronization problems between the time at which the compression rate is changed and the time at
which the compressed data is read out from the CT8022. When a 5.3 frame is read from the CT8022 Hardware
Receive Buffers after on-the-fly switching from the 6.3 rate, the Host must read 12 words instead of 10. In this case,
the last two words of data may be discarded.
Switching between the 6.3 and 5.3 rates is performed using the Select TrueSpeech Record Rate command (above).
To support on-the-fly switching, the CT8022 will accept this command after the start of the record process. If the
CT8022 is instructed to make an unsupported rate switch (for example, between TrueSpeech 6.3 and TrueSpeech
8.5), the status response from the Select TrueSpeech Record Rate command will be zero. The CT8022 record
operation will continue to operate with the old TrueSpeech rate.
The two least significant bits of the first data word of a G.723.1 6.3 or 5.3 speech frame are encoded to indicate the
speech frame type.
During playback (de-compression) the CT8022 will test the two least significant control bits only when the current
Host-selected TrueSpeech rate is 6.3 or 5.3. In this case, the CT8022 will automatically perform the correct
decoding of the speech frames independently of which rate the Host selected. To support on-the-fly switching for
playback, playback must be started with the 6.3 rate selected.
This fixes the frame size used for Host-to-CT8022 transfer at 12 words. When transferring a 5.3 speech frame into
the CT8022 when the 6.3 rate was initially selected, the Host should append two dummy words (zeros) at the end of
the 10-word TrueSpeech 5.3 frame.
To switch between other speech rates, the Host must stop and re-start playback/record. The Host is responsible for
synchronizing the rate switch between playback and record sources.
CT8022A11AQC FW Revision 0118 DSP GROUP, INC., 3120 SCOTT BOULEVARD
SANTA CLARA, CA 95054 PH: 408 986 – 4300 FAX: 408 986 – 4490
65
All specifications are subject to change without prior notice.
ETC [ ETC ]
