Getting sound to work in Linux

**AnonymousHero** · March 19, 2016, 08:52

I probably can't help with the concrete issue you're seeing, but if you could compile with debug information I'm sure a stack trace with line numbers would be appreciated

.

Adding "-g" to CFLAGS in the makefile or doing "CFLAGS='-g' ./configure" to configure should do the trick.

**calris** · March 19, 2016, 10:54

Ok, the plot thickens...

As far as I can tell, Angband only supports SDL for sound. SDL needs smpeg to process mp3 files, but smpeg is ancient and does not seem to agree with the latest build chains (see http://forums.fedoraforum.org/showthread.php?t=33457) and the last release was 14 years ago.

Now angband uses Mix_LoadMUS() for MP3 files and Mix_LoadWAV() for wave files. But Mix_LoadMUS() cane load wave files (see https://www.libsdl.org/projects/SDL_..._mixer_55.html)

So I tried using forcing angband to treat wav file like MP3 files, but now it seems to hit a problem in my_strimp

my_stricmp (s1=s1@entry=0x7fffffffce00 "summon_angel", s2=s2@entry=0x0) at z-util.c:92

So I guess my next question is rather obvious... does anyone have sound working in Angband on a reasonably recent Linux setup?

**Nick** · March 19, 2016, 11:25

Originally posted by calris

So I guess my next question is rather obvious... does anyone have sound working in Angband on a reasonably recent Linux setup?

Well, I certainly don't - I use OpenSUSE, and have had no success getting an appropriate mp3 player to work. I would be very interested to hear if people using other distros are successfully using sound.

Sadly, if you ask a Linux expert, they will typically say something like "Why not use ogg?", which is unhelpful in this circumstance.

I'm sorry I don't have anything helpful to contribute. I have made a few attempts to get this sorted out, but not made much progress. I am very happy to test any possible solution you might find, and to implement any actual solution.

**redlumf** · March 19, 2016, 12:54

Originally posted by calris

Ok, the plot thickens...
...

So I guess my next question is rather obvious... does anyone have sound working in Angband on a reasonably recent Linux setup?

I just downloaded 4.0.4 src. Sound works on kubuntu 14.04.4 LTS.

./configure --with-no-install --enable-sdl-mixer

./angband -mgcu -ssdl

Never had a problem with it.

**redlumf** · March 19, 2016, 13:36

Originally posted by Nick

...
Sadly, if you ask a Linux expert, they will typically say something like "Why not use ogg?", which is unhelpful in this circumstance.
...

Ok, go to your sounds directory.
Do this:

Code:

for f in *.mp3 ; do sox -v `sox "$f" -n stat -v 2>&1` "$f" `basename "$f" .mp3`.ogg; done

(btw, needs sox...install it if you don't have it)
If the above worked (ignore the warnings), you now have all your mp3 converted to ogg. (Also volume normalised cause I find the originals way to low on the volume side

Now, lets fix the sound.cfg

Code:

sed -i -- 's/.mp3/.ogg/g' sound.cfg

Done.

Enjoy your sounds in .ogg now.

**calris** · March 19, 2016, 14:21

Originally posted by redlumf

I just downloaded 4.0.4 src. Sound works on kubuntu 14.04.4 LTS.

./configure --with-no-install --enable-sdl-mixer

./angband -mgcu -ssdl

Never had a problem with it.

That is good news - at least there is no underlying bugs in the way the sound event system is set up. I was worried it might have suffered from bit-rot

I'm now working on a bit of a hack for raw ALSA support. It only supports a sample rate of 14400 and raw WAV files in the sounds directory at the moment.

It's been a long time since I've hacked Angband's source (way back in Ben's day - even before his massive overhaul). I was a bit worried about what I was getting into, but the sound subsystem is so elegent - it's a delight to play with the code

I've pulled out the parsing of the sound.cfg file and added callbacks into the sound module for loading of the actual sample files. Should make additional sound modules even easier to add

**calris** · March 19, 2016, 14:47

Originally posted by redlumf

Ok, go to your sounds directory.
Do this:

Code:

for f in *.mp3 ; do sox -v `sox "$f" -n stat -v 2>&1` "$f" `basename "$f" .mp3`.ogg; done

(btw, needs sox...install it if you don't have it)
If the above worked (ignore the warnings), you now have all your mp3 converted to ogg. (Also volume normalised cause I find the originals way to low on the volume side

Now, lets fix the sound.cfg

Code:

sed -i -- 's/.mp3/.ogg/g' sound.cfg

Done.

Enjoy your sounds in .ogg now.

You are a legend

- had to install "sox-plugins-freeworld" to get sox to know about mp3, but after that, Angband sound works perfectly. Thank you.

For those interested in the arcane, the SDL sound module processes anything NOT mp3 as a wave. So it plays mp3 files as 'Music' and everything else as a 'Sample'. No idea what the difference is under the SDL hood. So I'm a bit mystified as to why installing the .wav sound pack didn't work...

Now I'm off to finish my ALSA port

**Nick** · March 19, 2016, 22:15

Originally posted by redlumf

Ok, go to your sounds directory.
Do this:

Code:

for f in *.mp3 ; do sox -v `sox "$f" -n stat -v 2>&1` "$f" `basename "$f" .mp3`.ogg; done

(btw, needs sox...install it if you don't have it)
If the above worked (ignore the warnings), you now have all your mp3 converted to ogg. (Also volume normalised cause I find the originals way to low on the volume side

Now, lets fix the sound.cfg

Code:

sed -i -- 's/.mp3/.ogg/g' sound.cfg

Done.

Enjoy your sounds in .ogg now.

I guess I should have seen that coming

Let me put it in a different way.

The current Angband sounds are Dubtrain's sounds, converted to mp3 (because wavs were too large - 3M vs 35M). These play fine in the Windows and OS X ports, but only in some Linux distros, with the problem being scarcity of open source mp3-playing software.

Is anyone aware of a cross-platform way to solve all this, or do we just assume that Linux folk can sort it out for themselves as in this thread?

**takkaria** · March 19, 2016, 22:34

Originally posted by calris

I've pulled out the parsing of the sound.cfg file and added callbacks into the sound module for loading of the actual sample files. Should make additional sound modules even easier to add

This is excellent news. It's been needing doing for years and I'm glad you got there first

**Derakon** · March 19, 2016, 23:10

Is there any particular reason to use mp3 over ogg vorbis?

**HermannD** · March 19, 2016, 23:27

Originally posted by Derakon

Is there any particular reason to use mp3 over ogg vorbis?

Given how Nick described it, I would guess no easy support on Windows and Mac? If so, when was that last checked? Maybe that situation (or whatever the actual reason is) has changed.

**redlumf** · March 19, 2016, 23:34

Originally posted by Nick

are of a cross-platform way to solve all this, or do we just assume that Linux folk can sort it out for themselves as in this thread?

Isn't ogg a cross-platform solution? (You should have seen this one coming also :P)

**takkaria** · March 20, 2016, 00:12

Originally posted by redlumf

Isn't ogg a cross-platform solution? (You should have seen this one coming also :P)

MP3 is the most universal lossy format. You can play it without using an external library or with any extra downloads using Windows and OS X. That isn't the case for any of the Ogg codecs AFAICT.

**molybdenum** · March 20, 2016, 00:55

Originally posted by takkaria

MP3 is the most universal lossy format. You can play it without using an external library or with any extra downloads using Windows and OS X. That isn't the case for any of the Ogg codecs AFAICT.

OS X does not support Ogg containers nor any containers that can contain Vorbis audio. These are the native supported containers and data formats. There's not really much in that list that would be cross-platform, compressed, and provide good audio quality.

Getting sound to work in Linux

Getting sound to work in Linux

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment