Hi Anthony, > I am using a number of EPIA ML based systems,
the ML and MS are very similar > running Debian 3.1 (2.4 Kernel), same here, but with Kernel 2.6.14.3 + epia patches > and using ALSA (with PyAlsaAudio 0.2). still on pyalsaaudio-0.1 here > I am using the standard Shtoom codec (Mulaw as far as I know). > I am using OpenSER as a SIP registrar only, but this is running > on the same network (as all of the EPIA machines). my * is working as a media gateway too, meaning all my communication between the phones are bridged and transcoded if necessary on the *. Which should introduce more overhead than your OpenSER, as there (afaik there is no bridging) the two shtooms talk directly to each other. > Would I see any performance improvements by switching to > a codec such as Speex?What is the best codec to improve latency on > low-powered hardware? http://www.speex.org/comparison.html and I think that there is something on http://voip-info.org/ that does some comparison too. But I'm uncertain if it is a question of 'low-powered hardware'. You have the same effect with some other phone sw? And I would try a kernel with the epia patches, as I recall some issues with the ethernet module with a stock 2.6 kernel. Maybe with 2.4 too? Ciao, Frank _______________________________________________ Shtoom mailing list [email protected] http://mail.python.org/mailman/listinfo/shtoom
