[Asterisk-Users] Re: Why Asterisk can't cope with silence suppression?

[Peter Svensson]

On Wed, 16 Feb 2005, Rich Adamson wrote:

> > It's more than that, from what I know a *missing* RTP packet could be 
> > 'silence' (vad) or it could be 'network related' (jitter).  * not seeing 
> > a packet doesn't always mean it was vad, it might mean your network had 
> > a split second (subsecond) hiccup that caused the packet to disappear - 
> > both 'look the same' to *.  This is why someone had already mentioned 
> > the idea that the new jitter-buffer might handle this better/correctly.

VAD and lost packets do not look the same. When silence is detected the CN 
command is sent several times to ensure that it is not lost. A missing 
message should be interpreted as packet loss. CN is explicit.

> Personal opinion (and everyone's got one) is that vad does not produce the
> savings that one might expect. People are use to constantly talking (in
> many cases full-duplex-style), room background noise, dog barking, etc, 
> etc, which reducees the impact. Vad may have some small value for 
> residential voip users where their bandwidth is a little on the too-
> small side, but asterisk was not designed with the intent of putting a
> pbx in every home. 

Silence supression and VAD is not very significant on a connection that
carries only a single line. Most measurements I have seen in telecom
litterature quotes a typical ratio of 40-60% silence for conversations.  
When you have several calls active at the same time the savings add up. It
mostly affects big nodes with a lot of simultaneous calls.

Peter