Hi Arun, I've once done a simple test about how different resampler method could impact the CPU utilization. You can take a quick look. ;) (It's very high level, and it didn't include speex-float-1, maybe I'll try to add speex-float-1 and post the result again...) Generally, I don't have objections as long as the sound quality doesn't degrade too much, however, it might be subjective. On Wed, Sep 28, 2011 at 05:20:51PM +0800, Arun Raghavan wrote: > Hey folks, > We've seen discussions a number of times on IRC about this, and now that > 1.0's out the door, this might be a good time to consider this. > > The current default resampler is speex-float-3, which is on lower-end on > highly loaded systems can be problematic (cause an underrun flood, > f.ex.). The current choice among distributions seems to be between > speex-float-0 (~2.5-3x faster) and speex-float-1 (~2-2.5x faster). On > embedded systems people might plug in something lighter still (like the > ffmpeg resampler). > > Qualitatively speaking, I can't really perceive the difference between > speex-float-1 and speex-float-3. Might be a good idea to perform some > blind trials, though. > > So any objections to moving to speex-float-1 as the default? > > -- Arun > > _______________________________________________ > pulseaudio-discuss mailing list > pulseaudio-discuss at lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss -- guanqun -------------- next part -------------- A non-text attachment was scrubbed... Name: pa-resampler.pdf Type: application/pdf Size: 125073 bytes Desc: not available URL: <http://lists.freedesktop.org/archives/pulseaudio-discuss/attachments/20110929/dae0d24d/attachment-0001.pdf>
