Kaboodle/Noatun vs XMMS

There's a story (print version) on reviewing sound quality between Noatun, Kaboodle and XMMS. Not surprisingly, Noatun and Kaboodle come out on top, with the author stating that "I'd recommend the Noatun/Kaboodle solution to most audiophiles and people with decent sound gear." Unfortunately, the author didn't take the time to discover some of the other features and plugins that make Noatun a better all around player. An interesting, if audio-centric, read.

[Ed: Of course, this is really a review of aRts, KDE Multimedia and all those Open Source audio libraries involved. Noatun, Kaboodle and even XMMS win automatically.]

Dot Categories: 


by Iuri Fiedoruk (not verified)

For me, mostly all is resumed to plugins.
Can I play nsf (nes sound format), spc (snes sound), psf (playstation sound format) in noatun? No.
Can I get functionality of noatun with some xmms plugins? Yes.

So the choise is clear.

And besides, Noatun is slooooow.

by Funk (not verified)

Yes, there are slight differences in XMMS vs. Noatun mp3 decoding quality, Noatun does sound slightly better, but both of them have trouble recognising some broken? mp3 files, whilst the other works.
A good reference for this would be comparing to Winamp (it's had a lot of time to improve its' playback quality.. and does require a lot more cpu than XMMS for example), or the even better decoding quality of tools like Sound Forge and Cool edit pro.

But as an audiophile, what are you doing listening to mp3s?
ogg/vorbis being a 4th generation codec sounds a _lot_ better at a smaller filesize
and the decoder for it is built into libvorbis, so it oughta sound the same whichever player you choose. (there is no reason not to choose ogg/vorbis anymore since 99% of players support it)

and FLAC... well, it always sounds like it should, so hopefully all players will have flac playback possibilities by default soon. (it's how I store samplecds, my own material and anything else sensitive to the ear)

I still mainly use XMMS for the simple reason that many of the plugins I need for it aren't available for Noatun and hopefully this will be fixed in the near future.
(although, I think the Noatun playlist using a native gui is better than the XMMS counterpart)

Examples are the following:

Shoutcast broadcasting

Crystality plugin (to enhance the sound of mp3) and no, it's not trivial BS, most mp3 compression programs use low/high pass filters leaving a laughable frequency range for the end material.

dumb-xmms plugin (IMO the best sounding mod/s3m/xm/it playback library)

various plugins for snes/nes sound chip instruction playback (ie. authentic sounding console music)

JACK ( is the reason linux may one day become a viable solution for people in proffessional audio business, and hopefully arts will be dropped in favour of JACK or adopt the same kind of architecture (needing a total rewrite from scratch I presume)

Low latency is not a luxury, it's a requirement if anyone is going to ever be able to do any serious audio work on linux & variants.

JACK implements just this, and it is ready and more than good to go.

by Ac (not verified)

How can Noatun and Kaboodle produce "better" sound? If I play an MP3 in Noatun and XMMS, they both use the same MP3 decoding algorithm! They both send the decoded stream to the same audio device! So how can one player produce better sound than the other if both the source and output are the same?

by arts_sucks (not verified)

arts just sucks. it can't use hardware mixing, if i'm compiling something while playing an mp3, arts skips. playing a game, arts skips, dragging a window, arts skips.

its bloody useless.

- It doesn't support hardware mixing, even though a lot of soundcards do (mine does, when using ALSA)
- Even when I'm not playing *anything* at all, artsd still uses 1.5-2.5% CPU time /all the time/ on my Athlon XP 2100+ . That is *NOT* ok.
- Even when set to play on 'realtime priority' (in the control center), it skips all the time, e.g. when you switch to another window, or start a compile job. Yes, I do have a kernel with the realtime patch applied, and xmms does not have this problem.
-When it does mixing, CPU usage is even a lot worse. Note that this should not have to happen, because most soundcards have supported hardware mixing for a long time now...
- My pentium 2 333 MHz could play MP3's at 4-5% CPU usage, so you're not gonna tell me my XP 2100+ still needs 5% CPU power to play the same MP3, right? It should be more like 0.4% by now...

So these things basically render any KDE-based sound player unusable at the moment. I guess that's why everone is using XMMS (I know I am!)

I hope some KDE developers will read this thread, and acknowledge the need to take some action on the kde-multimedia front, esp. the aRts back-end :-)

Just did a little experiment to prove this problem:

- Running aRts, using ALSA for output, realtime priority, playing a 192 KBit MP3, takes about 4.5% CPU time constantly.

- Running XMMS with ALSA output, I often can not even find it in top, but when it's there it uses like 0.1-0.2% CPU time, as expected.

XMMS users should take note that turning of the option (in Preferences->Options) "Smooth title scroll" will save a lot of CPU cycles! I disabled the scrolling songtitle entirely (it's useless anyway), so now I just see the first ~ 30 chars of the songtitle. Disabling the equalizer visualisation helps, as well.

What I'd like to know is: what is aRtsd doing in these 4.5% CPU cycles anyway? I mean by simply looking at the numbers above it looks like it could decode the MP3 20 times over in that amount of time!

I agree completely. I have personally raised this problem on various forums already several years ago. Unless working with network transparency mode of ARTS ( what I never do) - I don't need it. What for ? Sound stream mixing? Aren't 32 streams in hardware enough for non-audio-professional home user?
Unified API - maybe. Network transparency - maybe. But give me hardware mixing.

by Christian Schug... (not verified)

Vorbis & MPC are the ways to go :-)

I don't like that artsd uses about 10 - 15% of my Athlon 900 and didn't even work 2 years ago when I started with Linux and had a p133.

Give me Noatun without artsd for OSS and espicially Alsa-modules.

I don't know what Noatun does with the soundfiles while it uses the processor that much.

And well, there's a arts-plugin for xmms. It's worth a test :-)


by Christian Schug... (not verified)

multimedia system for KDE:

At first I want to mention there might be some security issues but it's about usability:

Have a look where you're in contact with multimedia:

instant messaging
sound and video encoding, decoding, broadcasting
cd/dvd burning

Webbrowsing is quite ok but sometimes Konqueror crashes after clicking too fast, what isn't page related. And there seem to be some bugs when using the tree view (was moving files into subdirectories and suddenly Konqueror moved them into my home-dir what wasn't visible in that /LAN-tree).

Instant messaging: I use SIM and the original AOL-AIM for linux. Well, I need that different design to recognize the network and I don't mind three contact-lists for aol, icq and gabber.

If I watch a DVD or listen to vorbis files and want to discuss with my IMcontact about it, I would like to send a 100kbit-video-stream. I choose "send stream" and get a list of noatun(s) stopped with playlists full with mp3, vorbis, xvid, swf, cdda://, vcd:// , dvd:// ... and alternative buttons like "open file", dvd-browser, vcd-browser, cdda-browser [Perhaps the file isn't in that list or noatun isn't open].
Perhaps I get a new video-window or I have it docked into the wm-window (100kb streams aren't big), 3 lines for the dialog and my line to enter new text.
Some people would even like to encrypt everything :-)

Well, perhaps you don't want to send streams via IM but want to make a server.
By now I use icecast2 and darkice for vorbis and I have to disable ICQ-sounds that are send to /dev/dsp and will be broadcasted too. I have no clue about video.

I want to be able to mark some files in the playlist and toggle them on and off via shortcuts.

artsd is annoying, I can disable the xine border in icewm, noatun always has a border and a title bar in other windowmanagers. Looks very strange when using the k-jöfol-skin.

I still like to encode all files @ the commandline, it's fast and safe.
I don't like programms that use cdda to grab files and I don't want to lose the cdparanoia output. Sometimes I choose another drive to grab CDs with errors

CD-Text is essential, it should support cdrdao-toc- and cue-files.
I want to choose whether I decode oggfiles first and write them to disk or burn them on the fly.

Have fun

by Funk (not verified)

Don't get me wrong, ARTS is great work, and I have not experienced any trouble with it.
About the hardware mixing, I disagree, ARTS can guarantee a quality mix doing it in software, and hardware should be optional but not default since 90% of pc soundcards suck.
The cpu cycles used for this are trivial IMO.

But, arts is not good enough, it has the severe problem of extremely high latency, which makes it unsuitable for any kind of professional audio work.

Sign me up on the list of JACK instead ARTS

by Alex (not verified)

About the hardware mixing, I disagree, ARTS can guarantee a quality mix doing it in software, and hardware should be optional but not default since 90% of pc soundcards suck.
So define some configurable list of cards that do suck and cards that not suck, and make hardware mixing DEFAULT for cards that don't suck. I personally don't think that 32-channel mixing of SB Live sucks for non-professional / semi-professional home use. There is no reason to overload the CPU when there already is a dedicated chip for a particular task at hand.

by Funk (not verified)

So you'd want GFX cards to use 8-bit blenders for 32bit graphics as well then?
It's the same thing here, they offer hardware digital mixers with unacceptable algos (creative etc)
or in all too many cases, the mix enters the analogue domain (creative again)

If you think the cpu wins and sound quality loss gains are acceptable then fine.
32-64 channel mixing oughta be no problem, I used to do it on a 386 in dos (trackers)
so unless you've got some weird need of 2000 channel mixing, I don't see what the problem is.

Good cards:
M-Audio (Midiman)
and the list goes on...

by fault (not verified)

Uh, wtf? On modern SoundBlasters (basically, anything past the SB AWE), mixing is 100% digital and handled within the DSP directly. Naturally, sound that is from analog inputs, such as lien-in, microphone input, and cd audio would have to pass through a ADC. This is how it works on non-consumer oriented sound cards as well.

Most of the quality loss exhibited by modern SoundBlasters is the fact that it handles all wave audio at 48kHz. Most audio is NOT sampled at this rate, and thus must be resampled. However, for most consumer related, this isn't exactly a problem. It only becomes one if you are doing sound editing. I don't see many people who are using Noatun or xmms doing that :x. Anyway, this behavior is exhibited irregardless of if you are doing hardware or software mixing.

by Another Alex (not verified)

ir·re·gard·less Audio pronunciation of "irregardless" ( P ) Pronunciation Key (r-gärdls)
adv. Nonstandard


by Alan (not verified)

I agree with everybody else here, I like xmms better than noatun.

It starts up faster, runs better (uses less CPU), crashes less, looks better and it's easyier to use.

And I don't like artsd or esd, a decent soundcard does the same on hardware and works better.

It would be ok to use artsd or esd for network transparency (like mad), I don't know if artsd has this capability, but anyway, I don't need it.

by Stephan Sokolow (not verified)

I prefer Audacious Media Player. It's a second generation XMMS descendant via Beep Media Player and it's like XMMS and Beep combined without the bugs. for those who are interested.

I can't even consider things like JuK because they lack support for Amiga Modules (MOD, XM, IT, S3M, etc.), Chiptunes (SPC, GYM, HSC, NSF, etc.), MIDI, and other such esoteric formats.

It helps that Audacious has what is apparently the best-sounding MP3 decoder in existence. I also like the in-progress conversion to format detection by headers rather than extensions and the planned switch to LADSPA effects and libvisual visuals.