23.09.2014 00:25, [email protected] wrote:
On Mon, Sep 22, 2014 at 2:08 PM, Alexander E. Patrakov <[email protected] <mailto:[email protected]>> wrote: 22.09.2014 23:50, [email protected] <mailto:[email protected]> wrote: I'm bringing up an ALSA driver - 1c22400.iis-sgtl5000. This driver works fine with aplay. Something I'm doing in the driver is not making pulse happy. This is a trace with one terminal running 'pulseaudio' and a second doing 'paplay left.wav''. I don't hear anything when plays. If I don't load "load-module module-rtp-recv" I can play left.wav under pulse without problem. Does this speex resampler work on ARM? I have NEON FPU. Probably your speex has been compiled with --enable-fixed-point. Unfortunately, this does not work by default with PulseAudio 5.0 and its default speex-float-1 resampler. You either have to change the resampler manually to speex-fixed-1 in /etc/pulse/daemon.conf or apply this fix: http://cgit.freedesktop.org/__pulseaudio/pulseaudio/commit/?__id=__ac984f59d36ef555bc5b0df9af1cd4__8193d0d14f <http://cgit.freedesktop.org/pulseaudio/pulseaudio/commit/?id=ac984f59d36ef555bc5b0df9af1cd48193d0d14f> Thanks, that is the problem. I am using buildroot. What should tell the buildroot people to get this fixed? Or will it get fixed in the next pulse release?
It will be fixed in the next release, but, since it is not released yet, please tell buildroot people to apply this fix during the build. I am really sorry that there is no official stable branch.
Should I fix the speex compile in buildroot to remove --enable-fixed-point on FPU capable CPUs?
Sorry, I can't answer the question. But you have the CPU, so you can benchmark both ways (speex fixed-point + pulseaudio configured to use speex-fixed-1, speex floating-point + pulseaudio configured to use speex-float-1). A valid benchmark is to play a weird-rate (e.g. 16000 Hz) wav file with aplay and look at the CPU usage by PulseAudio. If the CPU is too powerful for you to notice the usage, try speex-float-3 or speex-fixed-3 :)
If you do perform this benchmark, please share the result here. -- Alexander E. Patrakov _______________________________________________ pulseaudio-discuss mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
