Best FLAC player for Slackware regarding sound quality
 

Hi there,

this is actually an update to an old thread about ripping CDs and FLAC audio player quality., but I didn't want to bump such an old thread.

I did another FLAC player shootout, and just want to share the results with you. The reason for doing it was the new VLC 2.0.5 package provided by Alien Bob (for info and download URLs see here). I had not seriously tried VLC in a while, because in my first shootout it just failed to provide acceptable audio quality with FLAC files. But I gave 2.0.5 a try --- and I couldn't be surprised more! Sound was excellent!

Recalling my earlier experience I didn't want to believe it, and started to do a few comparisons. But these initial experiments verified the superior sound quality, VLC now provides.

To make a long story short, these are the programs I tested with all kinds of music on good stereo gear (don't think, that all the differences could be noticed on low-end equipment...). I am only talking about audio quality with FLAC files, here, not other formats or the GUI or functionality are considered.

0. SoX (play)
This little, but incredibly powerful command line program simply called play, which is part of SoX, has nothing to do with the Sarbanes-Oxley Act. ;) Instead, it is still the benchmark in terms of audio quality with FLAC (and probably with any other format, as well, but I only tested it with FLAC). While the differences between VLC, the best of the GUI players, and Kaffeine are only noticeable in certain pieces of music and only with good stereo gear, you should be able to tell the difference between SoX and any of the others even with inexpensive equipment. It's like lifting a final curtain and removing the last barrier between you, the audience, and the artists performing the music, when you change from one of the other players to SoX. It's ranked 0., because it is out of competition, as the actual test was for focused on GUI players.
A minor annoyance were a few alsa buffer underruns with play, that caused drop-outs mostly at the beginning of a song. But this may be a configuration issue or related to the way the FLAC files were created (most of them were created during ripping by k3b).

1. VLC
In my first comparison of FLAC players VLC failed completely. While it was good with other formats, it was the worst of all FLAC players. Sound was flat, sound staging was poor, and it never made me move my feet, even with the funkiest music. All gone. The new VLC provides the highest resolution of all FLAC players I tried (maybe with the exception of play from the SoX package, but I haven't tested it this time). It provides excellent and stable sound staging, has the best control of large orchestras and singles out solo instruments with stable positions, so that you can tell, where all the instruments and musicians are in "your" room, even with complex arrangements. Also, the former softness is completely gone.
So VLC made it from last to first within a couple of releases. Chapeau to the VLC developers and thanks a lot to Alien Bob for maintaining and providing Slackware packages for it!

Not much to say about the other contenders, as nothing has changed significantly.

2. Kaffeine
It was my #2 overall, and is now #2 of the FLAC players with a GUI. Sound quality of Kaffeine (BTW, with all formats, not just FLAC) is still as good as it was, which is *very* good. Just VLC is slightly better, and to notice the difference you need to use really good stereo gear.

3. Amarok
Actually, this one gives me mixed feelings. It's audio quality is very good, actually, but it's noticably less loud than the others. Sound as such is very similar to Kaffeine, I wasn't able to distinguish between the two reliably with my eyes shut and the same music playing with the same volume.

4. KPlayer/MPlayer
Sound was a bit darker, less "open", less brilliant than with the above ones. People, who like to hear chorus music, will like just that, I guess. So, this might be more a matter of personal likings than one of quality, and the output of MPlayer is still quite realistic and natural.

No rank: Juk
Well, this one is better than expected. It is very dynamic and open, and could be first choice for rock and pop music. It's correctly described as a jukebox, because the music you will typically find in a jukebox sounds just good with it. Use it for your next party!
EDIT: Why no rank for JuK? Originally I had ranked it #5, but then I found that there was a problem with one of the cables. I'll have to re-do the tests for JuK. The results for the other programs remain valid, for the moment.

Functionality and Usability
During my test, one result from my earlier comparison was confirmed: Kaffeine remains my favourite player for everything, because it is the easiest to use, and does right, whatever it does. It is one of the best audio players (only second to VLC, now), and still the best application for watching tv via DVB-T on Linux, as, e. g., it has comfortable and fast channel scan that finds more channels than any other program available, and, like VLC, it doesn't require additional codecs. It lacks more advanced features of VLC, such as album covers or encoding capabilities, but the UI is user friendly and intuitive, and integrates well with the KDE desktop (works also fine with Xfce, of course).
Amarok is the most sophisticated application for playing audio files in my small selection. One feature I like is, that it can fetch and display song texts from the internet, and even scroll the text according to the progress of the song being played. In principal it is capable of playing music from network devices, but it failed to play audio files from my NAS. Most of the time, Amarok is a bit too much for my likings, but occasionally I use it, because of its features (and sound quality is very good, see above).
KPlayer has become part of stock Slackware since Slackware 14.0. While it is good to have a GUI for MPlayer that integrates nicely with KDE, KPlayer didn't convince me. Functionality of the GUI is rather limited. It should support DVB, but I haven't managed to get it working. In the past I had more success with UMPlayer and SMPlayer, both based on Qt, AFAIK, but event these didn't come close to Kaffeine.
Last, but not least, JuK is a good choice, if you want to organise playlists based on metadata instead of file hierarchies. The GUI is very friendly and intuitive, and does just what is needed. In terms of usability, it's on a par or even better than Kaffeine. It doesn't try to look as fancy as possible, but instead it avoids getting in your way. You want to listen to your favourite music, close your eyes and dream a little bit, right? So you need a GUI that allows you to point and click together your playlists, easily and quickly. And when the music plays, you don't care about the look of the GUI, right? If you answer "Yes", you will like the GUI design of JuK. Which is not to say, that it doesn't look good, it just doesn't have the bling bling of Amarok.

Equipment used for the tests
ESI Juli@ audio card (muted onboard sound, blacklisted snd_hda_intel) playing files from an external USB 3.0 hard disk as input device
Rotel RC-1082 Pre-amp
Rotel RC-1072 Power-amp
Canton Vento Reference 7 loudspeakers (first series)
AKG K701 stereo headphones
RME ADI-2 as headphone pre-amp, as the headphone jack of the otherwise excellent RC-1082 is not up to the task --- it't not good enough to hear the difference between VLC and Amarok clearly, which is quite noticeable with the RME ADI-2.

So much for now, take care, and all the best for 2013 to all of you!

gargamel

Interesting! If you get time I'd love to hear what you think about Audacious or xmms.

If you are planning to test more options, I'd love to hear about mpd:)

@D1ver. Xmms is good for FLAC. I haven't tested it under the exact same conditions, but I think it is as good as Amarok or even Kaffeine, but not quite as good as recent VLC (but definitely better than older VLC). Regarding Audacious, I haven't really tested it, also, under equal conditions, but after only a few minutes of listening I'd rank it #4 (same as KPlayer/MPlayer) or even below that.

But as I said: Even #4 or #5 in the ranking provide quite good audio quality. None of the players I tried was bad. Listening to FLAC audio files was fun with all of them. Only, a few of them were even better than the others, and VLC came out as the unique winner this time --- a big surprise for me, given my past experiences with it.


@sycamorex. Yeah, but not in the immediate future. Currently I like a simple setup, with direct access from the music player to audio files. mpd would require me to set up a client-server infrastructure. While this can be done on one machine, it wouldn't add any value. And if the mpd server would run on another machine, that machine would be required to run in addition to the device, the client is running on. Too complex for my current demand.
However, I have mpd on my radar as one candidate for a multiroom set up. But this will not be realized within the next few months. Then I will have a see, what streaming servers are available and capable of streaming FLAC. When I looked at the topic a while back, I found lots of servers for streaming MP3 (all), AAC (almost all), Ogg Vorbis (many). But servers capable of streaming lossless were rare.


gargamel

I have an extensive bootleg collection in flac, and I use mpd and the Sonata client as a local music player. You don't have to set it up as a network server.

Yes, I know, server and client can be on the same machine. But what's the advantage of such a setup compared to using players like Amarok? Why not directly access the audio files?

gargamel

All open source players use libFLAC, which reconstructs the PCM audio bitwise identical to the original recording, there should be no differences in decompression quality. Then these PCM bits go more or less unchanged to the audio driver. All players (esp. sox) have filters (for volume control, EQ etc.), but they can disabled.

For an impartial comparison I would record the player output using an ALSA or JACK loopback and then compare them bitwise. If there were any differences, I would fiddle with the settings, until they're gone.

For me it comes down to system resources. Media players like Amarok are memory hogs, while I get the exact same features using mpd and a good client for it, for a fraction of the resource usage.

For example, do I really need a MySQL database for a library? Why not just have it read from my directories, which are already organized, and save the library list to a simple txt file?

Apart from being light-weight compared to Amarok, mpd is IMHO a more convenient way of playing music while playing with your system. You can exit X, you can even log out of the system and your music still plays uninterrupted.

gargamel, I would suggest you try and educate yourself a bit about digital audio. As jtsn noted before me, there is no way you are hearing any differences, unless they these differences are caused by you (using different EQ/volume with different players) and not the software itself. Maybe you think you are hearing differences, but you aren't. That's called placebo effect. Please read on what that is and what a double-blind listening test is. Also, on how to conduct a proper ABX listening test to prove your claims.

I would also suggest to present your claims to a community that is specialized in digital audio, such as HydrogenAudio (http://www.hydrogenaudio.org/forums/). If you do, expect mayhem. Posts such as yours, with no proof whatsoever are considered completely worthless there (and I'm being very polite with that description compared to what you'll get there).

BTW, the HydrogenAudio knowledgebase is a valuable resource for anyone who wants to educate himself about audio quality and digital audio.

In short, you're not convincing me, or anyone else that has the slightest idea about digital audio with your claims. You may have fooled yourself with your results (deliberately or not), but you're not fooling anyone that really has a clue.

Sort of true, sort of not. If there are different backends being used (e.g. xine, gstreamer, vlc) etc. there can be a differences in sound quality.

No, as ultimately, everything uses libflac for the actual decoding.

I considered making this point, but then I realized that some of those players might have been using different versions of libflac.

Alien Bob's VLC build, for example, would be using a different version of libflac than the libflac you installed from SBo.

An interesting follow-up test would be to have those players convert FLACs to WAVs and then test the quality of the resulting WAVs. All libflac does is convert FLACs to WAVs. (What jtsn said).

The version of libflac doesn't matter, the format is lossless. Sure, newer versions might be (slightly) better at compressing, but decoded PCM audio will always be the same. From FLAC project's FAQ:

Sorry, not wanting to be harsh, but you are missing the point. Instead of reading a lot about theory, I prefer just listening to my favourite music. I am convinced that you would hear the differences, I describe, too, sane hearing and reasonably stereo gear provided.
Of course I know, what a double-blind test is, and I clearly stated, that I haven't executed a proper professional test. But I actually took care of obvious things like EQ/volume differences, see my original post: It became clear very quickly to me, that all the others are playing louder than Amarok. So I guess, that I was not fooled by that.
BTW, you can test this yourself, very easily. Neither a lot of time nor expensive equipment is needed for it. Even with medium-class PC loudspeakers you should be able to hear the difference. Do that test, before your reply, please!

Of course, the reasons for the differences I hear, may have to do with different parameter pre-sets. So, what? Does this mean, that there are no differences? If you were right, what were those parameters good for, if they didn't make a difference?

So all the players use the same library for decoding FLAC. And this means, they must sound the same? Really?
If you were right, why would the VLC developers have re-written their "re-samplers" (whatever that is), claiming, that they did so to improve audio output quality? This must be pure nonsense in your eyes, because, if I follow your arguments, improvements in audio quality are not possible, at all, because older and newer versions of VLC both rely on the same library for decoding of FLAC files, right? It could be fun for you to argue with the VLC developers and tell them, that they have seemingly no clue of digital audio... Finally, someone is here to tell them, what they've done wrong all the years, after 1 billion+ downloads... ;)

I suggest, that you just do what you know so much about: A double-blind test with the same candidates I used, and equally good gear. Sane hearing provided, you *will* hear the same differences like me. And, unlike me, you might be motivated to track down the root causes of the different sound of the various players. Please report back the results, here! But if you don't hear any differences, you should see your ear specialist. Have fun!


gargamel