It seems to be a fashionable to blog about experiences with PulseAudio, I thought I’d join in.
I’ve actually had some good experiences with PulseAudio, seeing some tangible benefits over the ALSA setup I was using before. I’ve got a cheapish surround sound speaker set connected to my desktop. While it gives pretty good sound when all the speakers are used together, it sounds like crap if only the front left/right speakers are used.
ALSA supports multi-channel audio with the motherboard’s sound card alright, but apps producing stereo sound would only play out of the front two speakers. There are some howtos on the internet for setting up a separate ALSA device that routes stereo audio to all the speakers in the right way, but that requires that I know in advance what sort of audio an application is going to generate: something like Totem could produce mono, stereo or surround output depending on the file I want to play. This is more effort than I was usually willing to do, so I ended up flicking a switch on the amplifier to duplicate the front left/right channels to the rear.
With PulseAudio, I just had to edit the /etc/pulse/daemon.conf file and set default-sample-channels to 6, and it took care of converting mono and stereo output from apps to play on all the speakers while still letting apps producing surround output play as expected. This means I automatically get the best result without any special effort on my part.
I’m not too worried that I had to tell PulseAudio how many speakers I had, since it is possible to plug in a number of speaker configurations and I don’t think the card is capable of sensing what has been attached (the manual documents manually selecting the speaker configuration in the Windows driver). It might be nice if there was a way to configure this through the GUI though.
I’m looking forward to trying the “flat volume” feature in future versions of PulseAudio, as it should get the best quality out of the sound hardware (if I understand things right, 50% volume with current PulseAudio releases means you only get 15 bits of quantisation on a 16-bit sound card). I just hope that it manages to cope with the mixers my sound card exports: one two-channel mixer for the front speakers, one two-channel mixer for the rear two speakers and two single channel mixers for the center and LFE channels.
Well, I’ve come across a couple of bugs where sound intermittently disappears when using PA, but setting to ALSA fixed those.
Other than that, I’m all for PA with the promise of being able to stream my audio to a pair of bluetooth speakers while simultaneously playing the audio out of a pair of wired speakers. Multi-room sound system here we come!
Haven’t tested yet tho, so it may be a pipe-dream 🙂
I too use PA and am happy with it but I would like to note that “edit the /etc/pulse/daemon.conf file and set default-sample-channels to 6” is considered special effort and not automatic by 99.99% of people on the earth, all of which would not be able to do it.
Rob: right. I think a GUI would help a lot here.
That said, I don’t think it is possible to automatically configure this with my sound card: it is possible to connect between 2 and 8 speakers, and there doesn’t seem to be any way to detect what the user has connected. If it was possible to automatically configure, the Windows drivers wouldn’t ask you how many speakers you had.
I just would like to note, that you will be able to reconfigure your speaker setup without edition daemon.conf.
That will happen in the near future with PA 0.9.15.
Sure there is no GUI for it today but maybe “tomorrow” 😉
PA will also automatically probe which combinations for playback and capture are working with your sound card and allow on-the-fly reconfiguration.
For more information on that topic you should take a look at Lennart’s blog.
On-the-fly Reconfiguration of Devices (aka “S/PDIF Support”)
http://0pointer.de/blog/projects/oh-nine-fifteen.html
Funnily enough I had the opposite experience. I have a similar feature exposed by alsa (the surround sound feature from the hda audio driver). It is done automatically on the hardware (just like what you’re doing in software here except there’s no CPu overhead) and exposed by alsamixer. In my case I couldn’t have this feature back when using pulseaudio even when changing the channels, only the two main speakers would play.