> > The basic question we ask is whether current mobile devices can compress > > and decompress voice data in realtime ... > > Depends on the CPU. With minimal complexity, the lowest speed for > real-time is about 100 MHz for a slow ARM core.what are you calling slow ARM core? I experimented voip with alaw and ?law on the ti omap 5912 board running at 192MHz and it basically sucks 80% of the cpu. As for speex i couldn't make it run correctly. I used gstreamer, so if you tell me you get good results on this kind of board with these codecs, maybe i have a major overhead in gstreamer which is something to investigate. I would appreciate if you could give me some insight. - Christophe> > Jean-Marc > _______________________________________________ > Speex-dev mailing list > Speex-dev@xiph.org > http://lists.xiph.org/mailman/listinfo/speex-dev >
> what are you calling slow ARM core? I experimented voip with alaw and > ?law on the ti omap 5912 board running at 192MHz and it basically > sucks 80% of the cpu.Obviously, the 80% CPU is not used for the u-law encoding. What I was talking about is Speex itself requiring about 90 MHz of a slow ARM core (can be less on a good core).> As for speex i couldn't make it run correctly. I > used gstreamer, so if you tell me you get good results on this kind of > board with these codecs, maybe i have a major overhead in gstreamer > which is something to investigate.That's possible. In any case, u-law conversion can be done with far less than 1 MHz... About Speex, you would likely need to enable ARM optimizations and set the complexity to 1 (default it 2). Jean-Marc
> That's possible. In any case, u-law conversion can be done with far less > than 1 MHz... About Speex, you would likely need to enable ARM > optimizations and set the complexity to 1 (default it 2).done with arm optimizations and i was still getting high load ... guess it's from gstreamer somewhere. I'll check that next week. thanks, - Christophe