Re-added everyone in CC (sorry for the duplicate) ----- Le 25 Avr 16, à 10:39, Thomas Petazzoni thomas.petazz...@free-electrons.com a écrit :
> Hello, > > On Mon, 25 Apr 2016 10:29:35 -0400 (EDT), Jérôme Oufella wrote: > >> I'm bumping that thread to ring@lists.savoirfairelinux.net for >> convenience. > > Thanks! I'm dropping buildroot@ from the list of Cc: since really this > is no longer Buildroot related anymore. I'm including a few other folks > from Free Electrons however :) > >> > Thanks again for this contribution. As a side note, we tried Ring at >> > work as a replacement for Google Hangouts, but unfortunately, the >> > quality level was really bad as soon as more than 3/4 persons joined >> > the call. Since I know Savoir Faire Linux is behind the Ring project, >> > do you know if this is something that might be improved in the future? >> >> When using Ring in a n-way conferencing setup, the peer >> initiating the conference bridge acts as a network and encoding >> hub for the rest of the audience. This means that if the >> conference hub either reaches bandwidth or cpu limits, the global >> call quality will inevitably be degraded severely, that would be >> a first point to verify. > > All of the people who participated to the testing were using a fiber > optic or cable connection, and at least I am sure that the one who > initiated the call was using a fiber optic connection (200 Mbit/s > download and upload speed), so I don't think bandwidth was really an > issue here. CPU might have been an issue though, depending on how much > CPU is needed by the call initiator when the number of participants > increases. > > Is it possible to install this "conference hub" on a server, so that > we could set it up on a beefy machine with a very good connection? > >> You may also be impacted by the ambient noise as I don't think >> Ring performs prioritization of the talker (correct me if I'm >> wrong) and the echo canceller may not be super efficient at >> times. > > IIRC, the issue wasn't really the audio quality in terms of ambient > noise or echo, but really the fact that audio packets were lost. Up to > 3-4 people, everything was working alright, it's really when we loaded > up to 7-8 people that it started to fail. > >> In any cases, any input from you that may allow us to pinpoint and >> avoid the quality drop will be of interest. > > Sure, will do! > > Thomas > -- > Thomas Petazzoni, CTO, Free Electrons > Embedded Linux, Kernel and Android engineering > http://free-electrons.com _______________________________________________ Ring mailing list Ring@lists.savoirfairelinux.net https://lists.savoirfairelinux.net/mailman/listinfo/ring