[sip-comm-dev] Buffer size of sending RTPManagers in SIP calls


#1

Hello all!

CallSessionImpl.initializeRtpManager sets a buffer size of
500ms on the
sending RTPManagers. This makes for a long latency, so I
wondered if there's
a reason for it. In my short tests, a buffer size of 125ms
provided good
audio quality with vastly reduced latency, but I fear that
there could be
side effects that I didn't find with my tests (like choppy
audio on slow
computers/different hardware?). Has anyone else experimented with the
settings?

As a followup to myself, I have to add that I just found out that I am
really changing the receive buffer, not the send buffer. Still it seems to
me that the buffer size can be decreased without degrading the quality even
to a value of 50ms, though I don't know if this is just for my special case.
Perhaps the default size could be changed in the sip-communicator trunk to
get some feedback from the nightly build testers?

My test environment:
Java 6
javasound data source
Windows XP on a fast PC
local network
ILBC as codec

Regards
Michael Koch


#2

Mi Michael,

Koch Michael wrote:

Hello all!

CallSessionImpl.initializeRtpManager sets a buffer size of
500ms on the
sending RTPManagers. This makes for a long latency, so I
wondered if there's
a reason for it. In my short tests, a buffer size of 125ms
provided good
audio quality with vastly reduced latency, but I fear that
there could be
side effects that I didn't find with my tests (like choppy
audio on slow
computers/different hardware?). Has anyone else experimented with the
settings?

As a followup to myself, I have to add that I just found out that I am
really changing the receive buffer, not the send buffer. Still it seems to
me that the buffer size can be decreased without degrading the quality even
to a value of 50ms, though I don't know if this is just for my special case.
Perhaps the default size could be changed in the sip-communicator trunk to
get some feedback from the nightly build testers?

I don't quite remember why we went for 500. I think that this is simply
what used to work best for my use case but we have been modifying the
media implementation a lot since then, so maybe things have changed.
I've set it to 100 in trunk so that we could do some testing.

Could you please also create an alpha3 issue for experimenting with
different values so that we look into this more profoundly during the
following months?

Cheers
Emil

ยทยทยท

My test environment:
Java 6
javasound data source
Windows XP on a fast PC
local network
ILBC as codec

Regards
Michael Koch

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@sip-communicator.dev.java.net
For additional commands, e-mail: dev-help@sip-communicator.dev.java.net


#3

Could you please also create an alpha3 issue for experimenting with
different values so that we look into this more profoundly during the
following months?

Done: https://sip-communicator.dev.java.net/issues/show_bug.cgi?id=408