Specifications

Voice Activity Detection Reference Manual (MGCP/NCS Version)
Mediatrix 4100
143
2. Define the jitter buffer length in the voiceIfTargetJitterBufferLength variable.
The adaptive jitter buffer attempts to hold packets to the target holding time. This is the minimum
delay the jitter buffer adds to the system. The target jitter buffer length is in ms and must be equal
to or smaller than the maximum jitter buffer.
Values range from 0 ms to 135 ms. The default value is 30 ms. You can change values by
increments of 1 ms, but Media5 recommends to use multiple of 5 ms.
It is best not to set target jitter values below the default value. Setting a target jitter buffer below 5 ms
could cause an error. Jitter buffer adaptation behaviour varies from one codec to another. See
“About Changing Jitter Buffer Values” on page 143 for more details.
3. Define the maximum jitter buffer length in the voiceIfMaxJitterBufferLength variable.
This is the maximum jitter the adaptive jitter buffer can handle. The jitter buffer length is in ms and
must be equal to or greater than the target jitter buffer.
Values range from 0 ms to 135 ms. The default value is 125 ms. You can change values by
increments of 1 ms, but Media5 recommends to use multiple of 5 ms.
The maximum jitter buffer value should be equal to the minimum jitter buffer value + 4 times the
ptime value. Let’s say for instance that:
• Minimum jitter buffer value is 30 ms
• Ptime value is 20 ms
The maximum jitter buffer value should be: 30 + 4x20 = 110 ms
See “
About Changing Jitter Buffer Values” on page 143 for more details.
4. Restart the Mediatrix 4100 so that the changes may take effect.
About Changing Jitter Buffer Values
Media5 recommends to avoid changing the target and maximum jitter buffer values unless experiencing or
strongly expecting one of the following symptoms:
 If the voice is scattered, try to increase the maximum jitter buffer value.
 If the delay in the voice path (end to end) is too long, you can lower the target jitter value, but
ONLY if the end-to-end delay measured matches the target jitter value.
For instance, if the target jitter value is 50 ms, the maximum jitter is 135 ms and the delay measured
is 130 ms, it would serve nothing to reduce the target jitter. However, if the target jitter value is
100 ms and the measured delay is between 100 ms and 110 ms, then you can lower the target jitter
from 100 ms to 30 ms.
Voice Activity Detection
The Voice Activity Detection (VAD) defines how the Mediatrix 4100 sends information pertaining to silence.
This allows the unit to detect when the user talks, thus avoiding to send silent RTP packets. This saves on
network resources. However, the VAD may affect packets that are not really silent (for instance, cut sounds
that are too low). The VAD can thus slightly affect the voice quality.
G.711 and G.726 VAD
The G.711/G.726 VAD is generic – when enabling VAD, G.711/G.726 sends speech frames only during
periods of audio activity. During silence periods, G.711/G.726 does not send speech frames, but it may send
Comfort Noise (CN) packets (payload 13) containing information about background noise.
Note: You cannot disable the G.723 VAD.
In the Unit Manager Network Administration Manual, refer to chapter Ports Parameters, section Port
Configuration Window.