Nortel Networks NN43001-563 - Manual (Page 69)

Dialing plans 69
Speech Activity Detection
Speech activity detection reduces the IP bandwidth used by typical
voice conversations. When Speech Activity Detection is enabled, no
voice samples are sent during periods of silence (from one side of the
conversation or the other). When a caller stops speaking, instead of a
"dead" line, the listener hears "comfort noise" generated to match the
previous background noise level when the caller was speaking.
Coders can send silence frames before the end of transmission during a
period of silence. Coders might omit sending audio signals during periods of
silence after sending a single frame of silence, or send silence background
fill frames, if these techniques are specified by the audio codec in use.
This background white noise keeps the telephone from sounding like the
line has gone dead - the listener can tell that the call is still up, and that
the person at the other end has merely stopped speaking. This technique
allows pauses during calls to sound almost the same as they would on a
standard telephone line. The primary benefit of Speech activity detection is
that it allows the IP Trunk to use bandwidth only when it needs to send voice
samples, thereby saving expensive WAN bandwidth for data traffic or other
voice and fax calls. Since normal telephone conversations include pauses,
and only one side is normally speaking, Speech activity detection reduces
the bandwidth used on a call by more than half.
For applications that send no packets during silence, the first packet after
a silence period is distinguished by setting a marker bit in the Real Time
Protocol (RTP) data header. Applications without Speech Activity Detection
set the bit to zero.
DTMF Through Dial
Preservation and transport of tones through the IP Trunk 3.01 (and later)
network is critical for Interactive Voice Response (IVR) services. IP Trunk
3.01 (and later) can be configured to ensure that DTMF tone information is
included in the packets that are sent through the IP Trunk 3.01 (and later)
network and that the tones are re-transmitted by the far-end gateway. The
duration information for DTMF signals is not transmitted; that is, long DTMF
bursts are reduced to a short standard duration.
Callers can access traditional Voice Mail or IVR services (for example,
"Press 1 for more information" or "Press 2 to be connected to our customer
service department"). Services that depend on long DTMF bursts cannot
be accessed.
In order to ensure that DTMF tones are being transmitted properly, the DSP
must be configured correctly in TM 3.1. If the IP Trunk 3.01 (and later)
node is configured to use a voice codec other than G.711, "DTMF Tone
Nortel Communication Server 1000
IP Trunk Fundamentals
NN43001-563 01.01 Standard
Release 5.0 30 May 2007
Copyright © 2007, Nortel Networks
.