[asterisk-dev] Noise generation

[Kevin P. Fleming]

Olle E. Johansson wrote:
> During my adventures, I have came across some issues related to comfort noise and noise generation. While googling, several patents related to noise generation shows up. Is this why we don't support it or is it beacuse we haven't got the code?

I don't think that's the case, no. The biggest issue in Asterisk related
to Voice Activity Detection is that nearly every part of Asterisk
(applications, channels, etc.) expect a continuous audio stream, without
gaps. There have been some efforts towards improving this, but there are
still significant architectural efforts required to make it work
properly. It's really a separate issue from CNG... that's more relevant
to endpoints that someone is listening to.

One of the first steps down this road will be to rewrite the way jitter
buffers are attached to channels: they need to be attached on the read
(incoming) side of network-related channels, and they need to be able to
be configured to *generate* packets on a regularly timed interval, so
that an application (or another bridged channel) that is expecting a
reliably timed, continuous audio stream can actually get it.

-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
skype: kpfleming | jabber: kpfleming at digium.com
Check us out at www.digium.com & www.asterisk.org