[Asterisk-Users] call center running Asterisk - sound quality-critical!

[Tamas]

Kevin P. Fleming wrote:
> Matt Roth wrote:
>
>   
>> These statements seem contradictory.  I know of no way (short of a
>> custom patch) to tell Monitor() to mix the in and out legs prior to
>> writing them to disk.  On the other hand, MixMonitor() does just that
>> and I believe it also buffers the writes in a way that circumvents the
>> I/O bottleneck associated with Monitor().
>>     
>
> Both of these statements are correct
It seems, MixMonitor is usable again (since yesterday's svn commit) so
it can save at least saving of one channel. We would test now MixMonitor
for this reason.

Kevin, does MixMonitor have buffering? How big is the buffer? Is it
possible to change the size? I guess, we are talking about buffering
voice samples and writing only a bulk of them to disk (e.g. in every 50
packets - 1second).
If there is no such buffer, do you think implementing it can be a real
solution? Storing into RAM needs too big RAM which can be a problem
(e.g. if we want to use monitor() for storing 60 concurrent calls for
min. 15-20 minutes).

Regards,
    T.