User Tools

Site Tools


wiki:music:taggingguide

Tagging Classical Music

1.0 Introduction

“Tagging” digital music files is the process of adding metadata to them -data which describes the music (who wrote it, who's singing it, when it was recorded and so on). Most music player software is written to be able to read these tags and use them to display pertinent information about whatever track is playing at the moment. For example:

In this screenshot, you can tell I'm listening to a part of Bellini's Norma because the music player is displaying the contents of the Album tag in the 'Album' column. You can also tell that it's the 35th track of that opera because of the contents of the Track Number tag. The 'Title' tag is displayed, too, so you know some of the lyrics that accompany that part of the opera. Basically, if you tag well, your music player makes it easy to find music to play and informs you in some detail about what you're listening to.

On the other hand, if you are careless about tagging, you might one day be told you're listening to Beighthoven's Fith Synfony; or that Callas is singing in an opera recorded at 1.54pm in March 1953 when Fred Simmerman was the recording engineer and Doris Filbert made the tea… Bad tagging -whether mis-spelled, over-detailed or just plain, factually wrong- is annoying and distracting in equal measure! So getting the right tags, typed properly, and knowing the difference between useful information and pointless gossip is important -and hard to do!

In particular, tagging classical music is harder than it is to tag any other form of music. I should know: I've been trying to get it right since I first ripped a CD in 1998! Unfortunately, most of the digital music world is designed by those whose musical tastes incline to the 'popular' rather than the 'serious' -and you can see that extremely clearly from the way you are 'expected' to tag music.

To take just one example: consult the ID3 'standard' for tagging music with its “genre”, specifically it's Appendix A, which is its list of “approved” musical genres. 125 different genres are listed there: 79 in the “official” tagging standard and a further 46 from the “Winamp extensions”. Thus, we have such genres as “Pop, Rap, Reggae, Techno, Industrial, Fusion, Punk, Ethnic” and so on: not being an expert, but that sounds like quite a comprehensive way to describe the diversity of today's 'popular' music. Meanwhile, we can tag classical music as merely “Classical” using the official tags, or as “Sonata” or “Symphony” if the Winamp extensions are allowed. All told, a grand total of just three possible genres to describe everything written from about 1200 to 2019 in the 'serious' mode!

Let us not even worry about why there are separate tags for “Composer”, “Artist”, “Original Artist” and “Album Artist”. You could argue that “Beethoven” is the composer of Symphony No. 6, and Herbert von Karajan is the artist (or maybe even the Album Artist)… but since most music players these days display “Artist” or “Album Artist” as the primary bit of information about a piece being played, if you stuck to that line of tagging, you'd end up listening to Georg Solti's Ring and Benjamin Britten's Brandenburg Concerti -which makes no sense at all to anyone that understands Wagner and Bach did all the heavy lifting for those pieces and Solti and Britten were merely 'conveyors' of the composer's intentions!

Getting tagging right for classical music is thus a specialism all of its own, and I'd like to spend the rest of this article explaining my approach to it; why I think these suggestions should be widely adopted by all those who listen to 'serious' music; and its benefits and pitfalls. Naturally, these are my ideas, not Holy Writ -but I hope to suggest that they are 'guidelines' or 'principles' that can usefully be followed by anyone dealing with cataloguing large quantities of 'serious' music.

2.0 Some Definitions

Before we go further, let's make sure we are clear about the meaning of some words which keep cropping up in the context of digital music and the tagging of digital music files.

NameDefinition
CDA physical, polycarbonate disk on which a digital recording of music is supplied. Could also be a DVD or other 'digital disk'
WorkA single, named composition by a composer. Can be comprised of multiple movements, songs, scenes, acts or other artistic subdivisions
AlbumThe name given to a work. Bears no relation to the CD, since a CD can contain anything from a part of a work to multiple independent works
Disk NumberThe sequential number of the disk being ripped from a boxed set. Would start at 1 and increments thereafter -except that we never use it, because we are concerned with 'works' not 'CDs'
TrackThe distinct 'units' into which CD producers split an album and which are separately readable from a CD by most players
TitleThe name of each track, usually consisting of the first words being sung in that track or of a music description of the work movement represented by that track (for example, “Allegro molto” or “Praise we great men!”)
ArtistThe composer of the work
Album ArtistNot needed, but the composer of the work if you really insist
Original ArtistNot needed, but the composer of the work if you really insist
ComposerThe composer of the work
SongSame thing as a track -so in this context, the first movement of a symphony can be a “song”!
Track NumberSequentially incrementing numerical identifiers applied to each track of a work when ripped. Thus, the first movement of a symphony will be track 1; the second, track 2 and so on. Note this isn't quite the same as “Track”: a CD may contain 15 tracks, but if the last three are a separate symphony, then the first of those three will be Track 13 according to the CD, but get ripped as track number 1.
CommentsFree-form text field in which details of conductor and performers may be recorded
YearThe year the recording of the work was begun
Encoded byA free-form text field in which details of the program used to rip or encode or otherwise process a track may optionally be stored
Album ArtA high-quality graphical representation of the artwork with which the CD was originally supplied
GenreOne of a relatively few descriptions of the general category into which a work may be considered to be a part. See Appendix A

3.0 General comments on Tagging

Tag data is the primary way we have of organising, finding and playing your music, physically and logically. It is important therefore that it is accurate, concise and consistent. Information that is not pertinent to your listening or searching functions is therefore to be avoided. So, for example, I would suggest that including the name of the recording engineer in any of these tags is inappropriate, as no-one of mortal woman born is ever going to think, “Today, I want to listen to all music that Alfred K. Smedley helped record”. There will thus be no need to ever search for “Alfred K. Smedley” -and thus including his name anywhere is extraneous to the functional purpose of the tags. Put another way, tags are metadata -data which describes the data we care about (which is the music). Tags which don't describe the music but which merely store data that no-one else will ever care about are just a waste of time and space, and should therefore be avoided.

Data needs to be managed strictly hierarchically to be useful. If we include a composer's name in a space that is usually used to store the year of recording (to take a silly example), that metadata is useless, as no-one will be expecting it to be residing in that location, so no program will ever be written allowing you to search the 'Year' field by Composer's name. Therefore, composer data goes only in the tags that we expect it to go in. Placing it correctly is what preserves its utility to us.

A logical consequence of this 'rule of utility' is: don't duplicate information you store in one tag in another. We have a track number tag to tell us that 'this is the third movement of a symphony'. Do not, therefore, type “III.” or “3.” in the track Title's tag. That would be another example of storing something in a place we're not expecting it to be stored, so it becomes non-useful.

Tag grammatically at all times. This is just another way of saying 'tag consistently'. If you call him Rachmaninov in one album's tagging and Rachmaninoff in another, you will store the one composer's works in two different locations and have to perform two different searches to find everything. This is less than optimal.

Use diacriticals where the native sense of the work requires it. Thus, it's Götterdämmerung, not Gotterdammerung and Leoš Janáček, not Leos Janacek. Do not, however, go overboard: if the 'native sense' means suddenly switching to using Cyrillic or Armenian charactersets and fonts, this is best avoided. For truly exotic languages (from an English speaker's perspective, at least), therefore, stick to English transliterations.

Tag data guides our physical storage of music files. That is, we should be able to read a tag and use its data to determine a physical placement on disk for that file. In this context, it's helpful to bear in mind that the physical storage structure we are aiming for is: composer → genre → album (i.e., work or composition). Thus “Benjamin Britten/Ballet/The Prince of the Pagodas (Knussen)” or “Ludwig van Beethoven/Orchestral/Symphony No. 6 (Karajan)”. With that in mind, it becomes more obvious why we keep the Artist, Genre and Album name tags clean, clear and free of detritus.

4.0 Specific Notes about Tags

4.1 The Album Tag

The Album tag is used to record a unique, short, generally-accepted name of a work, together with the surname of the 'distinguishing artist' which makes one recording of a work different from another in round brackets after the name. Thus, “Symphony No. 5 (Gardiner)” versus “Symphony No. 5 (Boult)”.

There is no need to include the composer's name in the Album tag, since that is the job of the Composer tag (and the Artist tag).

Do not mention the nickname of the work, if it has one (Haydn's Clock Symphony is actually Symphony No. 101, for example). The reason for not using nicknames is simply that their presence disrupts a natural sort order: Symphony No. 101 should come between numbers 100 and 102, but if words like 'The Clock' are in there, it perhaps won't. Besides that, nicknames are rarely assigned by the composer, and so are not 'authoritative' or meaningful in that respect.

Occasionally, the one conductor will conduct the same work in different years. In that case, it's acceptable to add the date of recording into the Album tag as well as the conductor's as the 'distinguishing artist' element. Thus “Symphony No. 9 (Boult - 1959)” and “Symphony No. 9 (Boult - 1969)” are sufficient to uniquely identify Adrian Boult's two recordings of Vaughan William's ninth symphony. If there is no need to distinguish by date to arrive at a unique Album name, don't include one (we have the Year tag for this information, after all).

Album names should not generally include musicological information such as key or time signature. “Symphony No. 101” does not need to be followed by “in D major”, for example, as no-one is ever likely to want to search for 'all music in D major today' on the one hand; and on the other, no-one will ever play Haydn's Symphony No. 101 in G minor, unless they had a particularly bad oboist doing the orchestral tune-up that day! The proper name, in other words, implies the key. It doesn't need to be spelled out.

It is acceptable to include cataloguing and numbering information, where this serves to make the Album name usefully unique. Thus Handel's “Concerto Grosso No. 2 (Brown)” is ambiguous, where “Concerto Grosso Op. 6 No. 2 (Brown)” is sufficient to distinguish it from “Concerto Grosso Op. 3 No. 2 (Brown)”. Where the opus number or other cataloguing information merely serves to clutter and demonstrate musicological brilliance, however, it is to be avoided. It is similarly acceptable to include a BWV number in the Album tag of a Bach cantata, or a K number in a work by Mozart or any other similar cataloguing scheme widely applied to the output of a composer: often these are the only realistic way to comprehend the enormous output of these composers.

4.2 Artist Tags

The artist tag is used to record the standard spelling of a composer's usual full name, without titles, awards or other biographical detail. Thus, it's Benjamin Britten (because although he was christened Edward Benjamin Britten, no-one ever called him 'Edward', least of all himself). It's also not “Lord Benjamin Britten” or “Benjamin Britten (1913-76)”. Similarly, “Edward Elgar” is fine, whereas “Sir Edward Elgar” is not.

Under no circumstances should you re-order the names of a composer to achieve a sort-of spine-of-an-encyclopedia feel. Thus “Britten, Benjamin” is unacceptable. We are tagging for files that are to be used on a computer: a computer is able to find and locate “Britten” no matter how it is stored within a tag, because that's what search engines do for us. We no longer have to mangle our data to make it retrievable.

These tagging guidelines do mean you have to find Rachmaninov's music under “S” (since his first name was “Sergei”), and Beethoven's under “L” (since he was Ludwig to his friends). But this is, at worst, a minor -and temporary- inconvenience. Knowing composers' full names is a way of helping to get to know and understand them a little better, after all; and your music player's search capability can always pick up the slack if you forget first names. After a while, however, you will barely remember the days when you didn't know that Poulenc's first name was Francis, or that Praetorius was a Michael.

Incidentally, unless you happen to be Russian, don't call Rachmaninov Рахманинов. English-language tags are fine. Also don't go overboard: yes, Rachmaninov is technically Sergei Vasilyevich Rachmaninov, but no-one this side of the Moscow Meridian is ever going to remember that middle (patronymic) name, so it's pednantic erudition, not helpful, to use it. Where there is a dispute about spelling or English transliteration (say, Rachmaninof v. Rachmaninoff v. Rachmaninov), you are more-or-less free to go with whichever version floats your boat -provided only that you are consistent about it. I'd suggest using the version shown on Last.FM, on the grounds that it will prevent some media managers/players telling you that you've tagged things incorrectly -but they have a tendency to refer to Արամ Խաչատրյան in the original Armenian when we all know and love “Aram Khachaturian”, so Last.FM are definitely not to be regarded as Gospel Truth on the matter. Use your own judgment, therefore, but be consistent.

Note that classical music has no 'compound artists'. That is, there is no classical equivalent of 'Simon & Garfunkel' or 'Hootie & The Blowfish'. Anyone about to suggest “Gilbert & Sullivan” needs to remember that it was Arthur Sullivan who wrote all the music for those operas. Since you're listening to music, it's Arthur Sullivan who gets all the credit.

For a starting list of 'acceptable' artist tags, see Appendix B.

Uniquely, we declare the composer (and artist) to be “Compilation” when a CD contains dozens of short tracks by different composers, none of which is signficant enough on its own to count as a separate 'album'. Usually, this is true of 'theme' or 'mood' music CDs, such as “Carols at King's this Christmas” or “Renaissance Wind Band Music from Spain”. Rather than tag up each track as a separate album by a composer you've never heard of and will likely never hear of again, just label the entire CD as one album by someone called “Compilation”. Since this breaks every other rule of the classical tagging game, I try to avoid these sorts of CDs! Sometimes, however, we don't get much choice in the matter :-(

4.3 Original Artist, Album Artist and Composer Tags

There are numerous tags used to record different data about non-classical music. Apparently, it's a 'thing' for one artist to “cover” (i.e., re-perform) a song originally sung by someone else. Thus, Kate Bush did a 'cover' of Elton John's Rocket Man. When tagging Kate's track, you (may) need to record the fact that the original artist for that song was Elton John. Hence Album Artist (Kate) will be different to Original Artist (Elton). Conceivably, too, you might want to record the fact that Elton John's music at one time was largely composed by Bernie Taupin -at which point, Album Artist=Elton, Composer=Bernie.

But we simply don't have this sort of chaos in the world of classical music. Bach wrote it, end of. The fact that Koopman, Suzuki and Gardiner have all had a go at conducting it is (almost) irrelevant: Bach is the Composer and the only person we really, truly care about. In the world of classical music, the Composer is King and everyone else are mere minions attempting to bring divine fire down from the mountain top.

Thus, for classical music tagging purposes, we merely assert that Composer=Artist: there is no distinction or differentiation to be made amongst either of these tags: the same data goes in them Both. On the same sort of note, Album Artist=Original Artist=nothing at all: we simply don't use these fields in Classical music tagging, as they refer to musical practices that just don't apply.

Mind you, since I've gone on at length about not duplicating information, you might very reasonably wonder why we bother to fill in even the Artist tag, then, since it only duplicates what is in the Composer one. It's a perfectly good question. Unfortunately, the answer is derived from the fact that the digital music industry doesn't really understand classical music. Take any music player you like and I'll make a small wager that it doesn't display the 'Composer' tag at all, or only if you dig down multiple layers of complication! It might be possible to re-configure it to do so, but out of the box, none of them will and most of them can't. You'll mostly find, instead, that it's the Artist tag which drives the display of most media players in existence, therefore it becomes necessary for classical music listeners to fill in this tag, even though for us 'Composer' is all that matters.

4.4 Disk Number Tags

Because our definition of an “Album” is explicitly not dependent on the physical CD a work happens to be supplied on, we have no concept of “disks” when tagging classical music. Götterdämmerung was supplied on 4 cds, but once ripped, we merely have a folder of about 54 tracks, none of which can thereafter to be said to have an associated 'disk'. This disk number tag is therefore always to be left empty.

4.5 Track Number Tags

Track numbers must be relative to the Album (which means the same thing as 'should be calculated per-composition').

In plain English, it means every “work” needs to be ripped independently of any other, no matter that they happened to be supplied on the same physical CD, and therefore each “work” must start with a track number 1.

If you have a CD which stores two symphonies, for example, then the second symphony will probably be physical tracks 5, 6, 7 and 8. When ripped, however, the second symphony must be tagged with track numbers 1, 2, 3 and 4. As a standalone work, its first movement deserves a 'track number 1' designation, regardless of what 'physical' tracks the thing consisted of when supplied on CD.

4.6 Comments

Comments are free-form text descriptions of the various performers of a work. They should list conductor, orchestra, chorus and principal soloists in that order, separated by commas. Thus: Georg Solti, Vienna Philharmonic, Vienna State Opera Chorus, Birgit Nilsson, Wolfgang Windgassen, Gottlob Frick, Gustav Neidlinger, Dietrich Fischer-Dieskau would be an example of a good comment tag.

It is also acceptable to indicate the function or role of principal soloists. For example: Georg Solti, Vienna Philharmonic, Vienna State Opera Chorus, Birgit Nilsson (Brünnhilde)… or Georg Solti, Vienna Philharmonic, Vienna State Opera Chorus, Birgit Nilsson (Soprano)….

Few players, if any, ever display the contents of the Comments field, so you are free to get as encyclopædic in your descriptions as you like -but it's generally not really necessary (since you can consult recording booklets or the Internet to fetch all the arcane details). I would therefore encourage short comments that contain the essential performer details and not much else.

4.7 Year

I would prefer to think of the Year tag as optional -but, unfortunately, there are a lot of music players out there which think it's important (or even entirely natural!) to organise music in date order. When the year tag is ommitted, such players simply lump completely disparate recordings together as essentially all coming from 'year 0', which is less than helpful. The fix for this is really to get a better music player/manager, but in the meantime, I would recommend filling in the year tag with an accurate record of the 4-digit year in which a recording was started. Thus, anything whose first recording day was on December 19th 1959 and whose last was on January 15th 1960 would be recorded as “1959”. Do not include months or days (or, perish the thought, clock times!)

Try not to use the year a particular recording was released, though. In the world of classical music, they regularly re-release the same recordings in different packaging, different combinations of other works, different giant boxed sets. If you use the date that particular boxed set was released, you'll be telling us more about your spending habits than the music!

In this context, for example, I have seen many times recordings with Benjamin Britten conducting dated “1986” or “1994”, which is a bit of an achievement since he died in 1976! Naturally, what was actually being recorded in those examples were the years Britten's 1968 recording of a work was re-released.

If you don't know or can't find out when a recording was actually made (some record labels' CD booklets are less than forthcoming on such details at times), you are allowed to make one up -to avoid the dreaded 'year 0' effect in some media players! I'd recommend trying to be as accurate as you can about it though. Use Allmusic or Discogs first to unearth detailed recording history about specific CDs if you can though, to avoid having to go the route of pure guesswork!

The metadata shouldn't look silly, basically.

4.8 Genre

There must be only one genre per work. The fact that you could theoretically characterise Beethoven's 9th Symphony as orchestral, vocal and choral doesn't mean you should! It's a symphony. Or it's primarily an orchestral work. Take your pick, but pick only one. Put another way: there is no sub-genre in the Genre tag, so please don't start doing things like “Symphony/Orchestral/Vocal”.

Useful genres are broad classifications, not precise descriptions of a specific musical form that a musicologist would approve of. Thus 'opera' is sufficient to describe Wagner's works; there is no need for a “music drama” genre, let alone a “Gesamtkunstwerk” one!

Genre is fundamentally quite a subjective way of defining music, so if you prefer to categorize piano works separately from harpsichord ones, rather than lump them together as 'keyboard', be my guest. But I would avoid too much specificity -so be prepared to call a clavichord a harpsichord for the purpose of genre tagging!

No matter that it's subjective, genre is important because it's one of the fundamental ways of physically storing music files. On hard disk, we should expect to find all of Bach's music in one place, all of Bartok's in another -so composer is the primary delimiter of storage. But within Bach's music, we probably would like to see his cantatas separately from his oratorios or his organ works. Thus genre becomes the secondary delimiter of storage. The 'album' (i.e., “work” or “composition”) is the third delimiter, so that we can use a file manager to navigate to Bach → Cantata → Cantata No. 140. Choose your genres carefully, therefore: sufficient to break up a large collection into meaningful 'chunks', but not so precise that there are a bazillion sub-divisions within one composer's output, which would be a management nightmare, not a convenience.

I have prepared a list of what I regard as 'useful' genres in Appendix A. I think they strike the right balance between fact and utility, without going bonkers-full-on-musicologist about it!

4.9 Title

Track titles should be reasonably short and either describe the music on that track in its composer's original musical instructions (eg, “Allegro molto”) or include the first few words of whatever song or aria or chorus is being sung on that track (eg, “On the beach at night, alone”). It is not helpful to include enormously long stretches of text: the Title tag is not the place to include the complete lyrics of an aria, for example. Musicalogical descriptions (such as Allegro con moto and so on) should obey grammatic rules and use normal case. Allegro Con Moto is not correct, for example. Neither is: On The Beach At Night Alone, as writing everything in InitCaps is ungrammatical English. Where German is involved, inappropriate capitalization is even worse, as capital letters serve to distinguish nouns within a sentence, so capitalising things which aren't nouns just renders the resulting German utterly nonsensical.

Do not include track numbers within a Title. We have the Track Number tag for that. Do not, therefore, tag the first movement of a string quartert as “I. Adagio” or “1. Adagio”.

Do not include character names or performance notes within the title tags, either. Thus “A Song for All Seas, All Ships: Behold, the sea itself (Chorus)” is not correct, as we can hear it's a chorus doing the singing and don't need the tag to tell us.

Similarly, “John Gielgud, Richard Pasco; Matthew Best: City Of London Sinfonia / Vaughan Williams: The Pilgrim's Progress - 01. Prologue” is all manner of wrong, including as it does the speakers, conductor, orchestra, composer, album name and pretty much everything else about the work within the track title. “Prologue” is all that's needed from this lot (remember: even the “01.” is wrong, as it duplicates the track number tag.

4.10 Album Art

The album art displayed when a media player is playing a track is important –it gives you a visual clue as to what piece is playing, for starters; additionally, it will often give you clues to the conductor, orchestra and other performers, without you having to shoehorn that information into inappropriate tag fields. Therefore, acquire your album art as large and as good quality as possible (within reason). 1000px x 1000px is a good starting point, though 500×500 is good, too. Anything less than that, however, generally ends up being too small to be really useful.

If you have ready access to both the original CD and a decent flat-bed scanner, scanning the actual CD jewel case booklet is a good idea -though trim it up afterwards and maybe re-scale it down to a useful 1000×1000 in the graphics editor of your choice. Failing that, do a Google search for the specific CD title (perhaps with performer names) and click on the 'Images' link when the results are displayed. You will often find useful album art versions for downloading there. Useful album art can also be obtained from Discogs and Allmusic websites.

In general, you should expect to use the 'real' album art -that is, whatever CD cover art was supplied with your actual CD of the music. However, I have bent this rule in my time. When I first got into the music of Benjamin Britten, for example, I bought lots of Decca LPs -and their artwork essentially became intertwined with the music for my brain thereafter. So, when Decca re-released all his music on CDs and decided to supply a variety of rather dull black-and-white photos of the composer as the covers for all of them, I decided to hunt down (via Google) examples of the original LP artwork with which I then tagged those CD rips.

Thus, although my CD of Britten's Rape of Lucretia was supplied with this art work:

…I preferred to tag my digital files of it with this:

…since that's cover of the LP version I first heard that opera on, and is therefore what I always associate in my head with that recording. It also happens to be the case that the older artwork contains a lot more useful information about the performers, orchestra and conductor than the newer CD cover.

The point is that the image displayed by your music player when you're listening to your music is actually rather more important than most people realise -so use whatever album art makes most sense to you in that context.

In all cases, whatever image you decide should represent the album art for a track, please embed the art within the digital music file, which essentially means using a third-party music tagger program such as The Classical CD Tagger to upload the album art in the form of a jpeg or png file into the track's own Album Art tag. Do not leave album art files sitting around on your file system as separate graphics files: make them become part of the music file itself. This will make the music file a bit larger than it otherwise would be, but the increase amounts to only a few Gigabytes on a huge music collection, so isn't really significant. Meanwhile, once it's part of the music file, your album art will always 'travel' with your music, without the danger of it being over-written or corrupted.

For the same reason, avoid the use of graphics files stored within the music folder called “folder.jpg”. Many music players/managers will read and display such files, sometimes even in preference to art that is embedded within a music file. Many will go one step further and update such 'folder.jpg' files without too much manual intervention on your part -at which point, your album art won't be anything that you recognise or see much meaning in.

So: the short version is, make your album art large, good quality, meaningful to you …and always embed it, rather than leaving it as discrete graphics files.

wiki/music/taggingguide.txt · Last modified: 2019/06/11 16:24 by dizwell