Fedora 12: Audio headache notes (ALSA and JACK)

Just a few notes about getting the sound working.

Pulseaudio

Pulseaudio belongs to the Linux Windowization era, meaning there are plenty of instructions of what-to-do-if but very little info about how the machinery works. Because it never fails, right? Who needs that info?

My notes about problems with sound from Firefox are in a different post.

Important: Any user needing to access Pulseaudio sound must belong to the “audio” group, since the device files in /dev/snd/ need to be opened.

Anyhow, if this is what you get:

$ aplay -D hw:0,0 file.wav
aplay: main:654: audio open error: Device or resource busy

Then just go

$ killall -9 pulseaudio

or

$ pulseaudio -k

(which is a pleasing action for some reason) and the pulseaudio daemon will (unfortunately…!) restart as a result of some other will-always-work hocus-pocus daemon, whose identity I may still need to find (rtkit?).

To prevent pulseaudio from restarting all the time, edit /etc/pulse/client.conf and change “autospawn = yes” to “autospawn = no” (possibly uncomment the line if necessary). Note that in the absence of the daemon, the volume control applet goes away.

To start it again, go

$ pulseaudio -D

Update: There’s the pacmd command-line utility, which supplies pretty useful info about what’s going on internally. If it says “No PulseAudio daemon running, or not running as session daemon” just kill -9 pulseaudio. That seems to be the solution for everything.

Using aplay and lsof, I managed to figure out that hw0,1 (Card 0, Device 1) is in fact /dev/snd/pcmC0D1p. To check who’s using my default sound device, go

$ lsof /dev/snd/pcmC0D0p

Pulseaudio may be in the list, because it’s running the device. When running aplay, it takes the file. Sometimes nobody does. I’m not sure what’s going on here.

And as silly as this comment is: To change the output sink, just use Gnome’s Preferences > Sound and set it up there. Also try “pavucontrol” from command line for a slightly different GUI tool.

ALSA

For (too much) information about the audio hardware, run alsa-info at command prompt.

The devices on my system:

$ aplay -l
**** List of PLAYBACK Hardware Devices ****
card 0: Intel [HDA Intel], device 0: ALC888 Analog [ALC888 Analog]
  Subdevices: 0/1
  Subdevice #0: subdevice #0
card 0: Intel [HDA Intel], device 1: ALC888 Digital [ALC888 Digital]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: HDMI [HDA ATI HDMI], device 3: ATI HDMI [ATI HDMI]
  Subdevices: 1/1
  Subdevice #0: subdevice #0

Jack

The idea about Jack is to allow the user route sound sources to sinks, so there is better control of what goes where. This makes sense in particular when MIDI devices are involved, since picking the MIDI player can make a significant change.

To work correctly, Jack relies on the audio group, which the user of the Jack application must belong to. Prepare for a relogin when fiddling with this.

Applications requiring JACK must have the Jack server running before being executed. QJackCtl is the preferred tool for that. Settings: Input/output device is hw:0,0 and sample rate 48000. Then press “Start” to kickoff the server.

Jack needs to be exclusive on the hardware device, so if another application is using hw:0,0 jackd will not start, resulting in annoying periodic popup messages by QJackCtl. To find out who’s to blame, see notes about Pulseaudio above.

Also, if it manages to start, no other sound application (including console alerts) will be able to play any sound. Did I say annoying? Or hiJack?

This is what happens when jackd is running, making all other sounds die out.

$ lsof /dev/snd/pcmC0D0p
COMMAND   PID USER   FD   TYPE DEVICE SIZE/OFF  NODE NAME
jackd   14825  eli  mem    CHR  116,7          14591 /dev/snd/pcmC0D0p
jackd   14825  eli   10u   CHR  116,7      0t0 14591 /dev/snd/pcmC0D0p

And finally, this is the setup screen that worked for me:

Audacity

Can have its audio devices set to “default”. hw:0,0 has been used in the past to solve noisy sound problems, but it looks like killing pulseaudio is much better (and more fun).

LMMS

The GUI looks a bit like those cars with ultraviolet lights underneath. More like a toy, and not a professional tool. But judging from its variety of sound sources (some of which are outright stupid. Commodore 64 emulation? Come on) and its by-programmers-for-programmers feel I gave it a go. So it’s indeed mostly a nice toy: A fantastic variety of sounds, fun editing, but if you want to insert a few bars in the middle of an edited song, there’s no way to just make some space. No bulk operations, as needed when you really want to do something serious. So as for v0.48 it’s nice to play with but not really a working horse. The real-life production scenarios are not handled yet (in particular, moving and copying large portions). The undo operation is also very partially supported, so a simple mouse click can easily destroy some work irrecoverably. Having said all this, I hope its developers will stick to it, and I will take all my criticism back in a year or so. It looks like they are on the right path.

I was surprised to find out that the output format is binary, but that’s just compressed XML. If hand editing is desired, it’s possible to save to .mpp rather than .mppz.

Ah, and using ZynAddSubFX as a sound bank, I ran into a bug causing notes to be stuck playing indefinitely during playback and conversion to WAV. This happens at random. Just some note refuses to release until reloading the entire project. Making WAVs is impossible with this bug, of course. But it looks like I know how to fix this.

Rosegarden

It’s maybe a wonderful application, but I have to admit that its dependency on Jack made it far less attractive compared with LMMS. Having all pulseaudio dependent applications die out is not fun. The alternative would be to connect all other applications to Jack, which I would maybe consider had I been a real sound guy.