Which players for classical music

Hi everyone.

I have few cd and long time ago I convert them to some wav then aiff file, at that time I’m using iTune to «tag» them.
To listen them I’m using a server with MPD and event it’s not perfect that ok for me.
Recently I do a resampling the all collection of wav file to 192khz and convert them to flac (only because wav don’t have metadata). So now I’m re-input all the metadata to those flac file.
I try to follow the musicbrainz advice for album, title, etc…and find that actual not usable.

Why ? Let’s say I have 10 differents version of «Symphony No. 6 in F major op. 68 “Pastoral”» with different conductor, ensemble and year. If I follow or just extract the metadata from musicbrainz, I’m absolutly unable to use mpd/ncmpc/cantata/ncmpcpp/etc… because those application cannot use all the metadata and don’t have the level of flexibility to choose what I want.
So currently I manually fill the metadata to my 600Go for file so the album would be the «Symphony No. 6 in F major op. 68 “Pastoral”» for the 10 versions, the title of the track would be just «Allegro ma non troppo» , conductor would be «Wilhelm Furtwängler», Artist would be «Wiener Symphony Orchestra». With those information, I am able to list all my Symphony No. 6 very easily then I can choose the conductor etc…
So because the size of musicbrainz database I’m guessing I should be crazy or missing some part of my brain, I can see how you use to play the music.
From my point of vue the guideline of musicbrainz are focusing on playing by CD (just like old time). Meaning you browse the list of «Album» who can contain the symphony No. 6 and a symphony from Brahms and select the track of the symphony No. 6.
This default of this approach IMHO are you need to know where are each piece on each CD. For example I’ve got the box of 100 cd with all work of Beethoven, with the physical box I can find quite easily where are the symphony No. 6 but with the virtual version and the way musicbrainz put the metadata I find that very difficult to do.
I think I’m missing a big point somewhere…so my question are, what player you’re using to play your classical music.

I have a collection of classical music of similar size, so I’m familiar with this problem. As far as I know, the easiest way to do something similar to what you do would be to use some music collection manager on one side to order your music and then whatever music player you like to actually play it. Apparently instead of the (easier) ordering by albumartist/album you are looking at something like composer/work/performer. For that I see two possibilities: either the Classical extras plugin (Classical Extras 2.0) on Picard (the official musicbrainz tagger) which I never used but if it works as advertised it should do this. It will crash if you want to feed him 600 GB at once but if you do it in smaller bits it works. The other possibility would be to use beets (https://beets.readthedocs.io/en/stable/) which is a command-line music tagger that uses musicbrainz data together with the parentwork plugin (that’s what I use so if you need help on it you can ask me).
The first has a graphical user interface, the other is pretty much command-line only. The first orders the music once and if the metadata changes, you need to reiterate the process (unless I missed something) while the other has an infrastructure to keep track of changes on the database.

1 Like

Thanks for your answer.
Well I already try the Classical extras plugin. And yes he does a lot of thing, but the point are I didn’t find any «music collection manager», mpd/ncmpc, kido, vlc, audacity who was able to exploit those metadata.
Most of the software I encounter are designed for pop/modern music so juste artist and title are enough (ok it’s a little exaggerated…:wink: )
Currently I already does half of my CD so I will finish it with my rule of tags. But event then I really like to exploit the other tags.
Just thinking of another «technical» problem with musicbrainz recommanded way to put metadata are the lenght of the title. Seem stupid but «Beethoven, Symphony No. 6 in F major Op. 68 «Pastoral», Wilhelm Furtwängler, Wiener Symphonic Orchestra, 1950, Allegro ma troppo» are quite long. And I didn’t want take the example with a symphony with singer (like the No. 9 or some by Mahler) the list of the performer put in the title make the title so long no software will show, or you need to wait 30 sec.

So all I saying are, I don’t think the problem come from how musicbrainz tag but from the player, that say, because musicbrainz contain lot of classic disc I’m really wonder how people use it…with which player.

What «music collection manager» are you using ?


Yes on windows long path names can be an issue. As I said, I have on one side a music collection manager (beets) and on the other side players that I just feed the directory I want to play (the directory being of the form composer/work/performer/album/track or albumartist/album/track , I have both linked with softlinks). That’s the best I could come up so far and I don’t expect any better, because while most music file formats have a field ‘work’ it usually contains something useless like ‘Symphony no. 6: I. Allegro’ where what I would like to look for is ‘Symphony no. 6’.

1 Like

Ah yes…that’s f*g good idea :wink: :wink: Thanks.

I will write some code to convert from metadata to something on the filesystem through symlink (I’m under linux).

Thanks a lot

If you use beets, there’s already a plugin for that.

If I understand the question, it is about the music collection manager rather than the player. I have used Muso for many years and I still rate it very highly. See http://klarita.net/muso.html. I designed Classical Extras to work closely with it and I am very pleased with the result. You can see more details about how they work together here: http://music.highmossergate.co.uk/

1 Like

Thanks. But I running under linux and mpd.

iTunes (which is called “Music” now) can handle all the relevant tags for classical music just fine, it uses “work”, “movement” and “movement number” and shows it in a pleasant manner – you can even have the booklet (as a .pdf-file) stored together with the album. Maybe you give it a second chance?

Well…to Music maybe, but to a macintosh…no way ;-).
Just to be clear, It’s not against Apple (well…:wink: ) but because of my audio system work with linux and also because I’ve some knowledge in linux, so I can tune my linux much more than a apple computer. Like disable almost everything so I get the best sound I can.
I know that sound crazy but on my system wave are a little better than flac…
I didn’t try against a Macos but against a windows, my linux+MPD sound much much better than a windows 10 with the same files.

I’m not an Apple evangelist (well … :wink:) and everyone is entitled to go his/her own ways and to believe whatever they want. (Like that they could hear the difference between wave and flac – which is impossible, because both deliver the same bits to your audio system, which is: wave, uncompressed audio. flac is just for storage, it will be always decompressed bit for bit into the original wave for the purpose of listening.) So, yes, it sounds crazy …

1 Like

Yes I know it’s sound crazy, I’m work with computer, I’ve a some knowledge in mathematics too, first I don’t believe it when some member of my familly tell me they are a difference. I say exactly the same thing as you say, because flac are a looseless compression method so mathematically it’s impossible they are a difference. But… they are. We also invited some friend working in the high end audio, and he can also notice the difference. I didn’t say the difference are big, and I didn’t say you can notice the difference on a iPod, Smartphone event with some very good headphone. But with some crazy audio system it’s noticeable.
Since then I try to know where the difference can come. The most plausible explanation are when you use flac the cpu work a little more so the interference are little more. I event buy some very powerfull CPU, but that’s not solve the problem (the linux so almost no activity on the cpu). But currently I don’t know if it’s the actual reason.
After saying that I’m still using flac because wave without metadata are so f*cking nightmare to use…
And I will definitively say the oversize of wave file do not worth the very small gain in quality (and I’m little crazy on that point).

Did you make a blind ABX-Test (Wikipedia)? Probably not. If not, than you can never be sure if you are tricked by your own biases. And please NEVER make an ABX-Test with lossy files like 256 kbps AAC or 320 kbps mp3 vs. wave – you might find out that you even can’t hear that difference (like 99,99999% of all humans) and that your ears aren’t as exceptional as you currently think (and that your expensive audio equipment may have been a huge waste of money) … :crazy_face:

1 Like

Yes the test we do are blind test shuffle with rand function on linux. And we don’t have any lossy files.
And don’t worry my expensive equipment are all home made almost from raw material. Event the most of the transformer are home reel up. I didn’t say it’s cheap but compare to 99.99999% of the high-end equipement it’s super cheap and work better than 99.999999% of the expensive equipment and work better than 100% of the super expensive equipment (In fact I never find those super expensive equipment good in anything else other than prove to the buyer «I’m successful man/women»).


1 Like

An ABX blind test is nothing self organized on your couch with some relatives and neighbors where everyone says his/her impressions. An ABX-Test is a software (you can find it for free online, for example “foobar 2000” has it built in or as a plugin, I can’t remember at the moment). It plays the same piece of music several times and you only can click “A” or “B”, not if one is lossy and the other lossless, just “A” or “B”. At the end there is a statistic that will be 50% A and 50% B as the randomness would suggest. This means that you CANNOT distinguish between lossy or lossless. ONLY when you have a significantly higher percentage on one side, like 80% – 20% you proofed that you CAN distinguish. You should make a bunch of tests in that way and increase the bitrate of the lossy file gradually until you get 50% – 50% – that is the lowest bitrate over which you cannot hear any difference. Most people (even self-proclaimed “audiophiles”) fail above 192 kbps to hear any difference. Everything above that is a waste of storage and money. I bet you WILL NOT be better than 192 kbps, regardless of the codec. Let alone you would hear the difference between wave and flac, which you could test as well, if you want.

I’m not sure what you mean by “cannot”: mpd recognizes quite a lot a musicbrainz specific tags (https://www.musicpd.org/doc/html/protocol.html#tags) as long as you add them to your metadata_to_use option. I can display most of the information I am interested in with ncmpcpp for example so browsing classical recordings works fine for me.

(only 10? pff, amateur :stuck_out_tongue: )
I admit I’m not using the search very often in ncmpcpp, but I had the work MBID tag added a few years ago for exactly that purpose and it seems to work correctly (ok, looking for an MBID to paste in the ncmpcpp search engine is not exactly fast).

I can see that the “work”, “conductor” and “grouping” tags will be added to the next MPD version also, so that might help.

For the record, foobar2000 uses WAV’s RIFF header to store metadata successfully.

Unrelated but something I love is that it uses CDDB file system to allow tagging of Audio CD themselves directly too.
But I too have to find a nice Linux player now that I dropped Windows XP.

And I’m 100% sure you did that between a uncompress flac and a ultra-compress mp3 before you can say the uncompress flac sound better than the ultra-compressed mp3 ?
But don’t worry as you say it before

everyone is entitled to go his/her own ways and to believe whatever they want

:crazy_face: :crazy_face:

Just let me believe in the fact I can notice a difference :rofl: :rofl: :rofl:

Maybe I’m misusing something somewhere. It’s also the reason I ask… :wink:

I will stick with the example of Symphony No. 6, if I got a version from a box of CD (historical recording blablabla) the album tag (from musicbrainz) would be «Historical recording blablabla», so I will find the Symphony No. 6 information in the title, GROUPEHEADING,WORK, TOP_WORK,STYLE,GROUPING. Beside the tilte, none are really supported by mpd.
For the title it’s just a nightmare because it’s not grouping enough (think of Swan Lake, or some opera who got 40 tracks).
I will love to not have enter information manually…
Currently my workflow are get metadata from musicbrainz, an replace manually the album by the name of the work (Symphony No. 6).

And yessss i’m waiting for the support of conductor…

I didn’t know how to use MBID in ncmpcpp I will try to find out.

thanks for the help

Then I guess you need to wait for the next version of MPD :slight_smile: (or compile it yourself).
You could also use the undefined tags in MPD (name, comment) to store some of these fields until then.

What I meant here is that the “Any:” search box in the search engine looks in all known tags, so if your tag musicbrainz_workid is filled you can look for e.g. “Any: d4b8437e-5ee6-33b2-9462-3bb0d51ec0a0”

1 Like