The Axioms of Classical Music Tagging

1.0 Introduction

How one goes about organising a large collection of digital classical music is a problem that has many potential solutions. Here, I offer you some guidelines as to how I think you might go about creating a well-organised, easily-searched, efficient and effective classical music library. I will say right up front that whilst my suggestions will work on every media player I’ve tested on Windows, Linux, Solaris and Raspberry Pi, I haven’t tested them on Apple hardware (though I did run iTunes on Windows). If you are an iTunes user, my strong advice is to stop being one and find a different media player instead (one that has a development future would be a good start!) If you nevertheless wish to continue reading as an iTunes user (though probably one now a little bit insulted!), be assured that the tagging strategy I’m about to propose will work on that software.

Fundamentally, efficiency and effectiveness in this context comes down to how you tag your digital music files: that’s the business of adding metadata to your music files. Metadata is simply ‘data about data’: in this case, the data you care about is the music, and the data ‘about that data’ is textual information you add to the file to explain various things about that music. You might, for example, want to say who wrote the music, who is performing it, when it was recorded and so on.

The fundamental problem I see with classical music fans when they start tagging their music files is that they approach things in the way you would a library card index, or a set of books on a shelf. For example, you will see perfectly sane and sensible people suggesting that you should tag the composer field as (for example) Bach, J.S. or Britten, Benjamin. They do this, I think, because that’s what you would see in an encyclopedia entry for those composers; it’s what you would see in an old-fashioned card index in a library. It’s an approach that is useful in those contexts, because when you have to flick through the thousands of pages in the New Grove Dictionary of Music and Musicians, you probably know a composer’s surname, but not always his first (quick test: what was the first name of Biber? [and I don’t mean Justin!!] or Mouton? See…)

The trouble is, however, that we listen to digital music in the context of a digital world -in which we have computers that can sort and search by anything stored anywhere. There’s no need for us to mangle our metadata into unnatural word orders for us to be able to make effective use of it! Store your composer names as “Britten, Benjamin” or “Benjamin Britten” and the computer will find the relevant music with a search for ‘Britten’ in either case. Digitisation means we have the luxury, at last, of making our metadata look natural and attractive, rather than revert to artificial and unnatural forms that were previously mandated by the need to perform manual searches.

Another common problem I see when classical music fans tag their digital files: they do so with what I call the LP mindset. That is, in ye ancient days, you bought your music on LPs, which you spent quite large sums on, played carefully on good-quality kit (or wished it was good quality if it wasn’t!), and the process of playing the LP was part of the whole listening experience. The ‘LP mindset’ is the hangover from those days, when the LP was the thing you acquired and played and looked after. You didn’t have a performance of Beethoven’s fifth symphony on your record shelves; you had Karajan’s Beethoven 5, or Solti’s, or Mehta’s. If it was combined with a recording of Beethoven’s 6th as well, so much the better: an LP that had different compositions on its two sides was better value than one that only contained music from one. And this ‘LP mindset’ has been transferred into the digital realm, too. These days, we mostly play CDs of music -but still the idea that the CD is the object deserving attention and care persists.

Practically, this means that a lot of people tend to tag their albums as ‘Karajan – Beethoven 5 & 6’, because both works come on the one CD and it’s the CD you’re ripping and cataloguing.

But I want to urge you to forget this approach. It fundamentally misunderstands what a CD is (or an LP was, for that matter). A CD is merely a physical medium that is used to ship music to you. It is equivalent to the envelope within which your aunt pops your annual birthday card -and has about as much importance to things, too. Just as it’s the card that goes on the shelf, not the envelope; so the CD is irrelevant to cataloguing (or ought to be). Who cares that Beethoven’s 5th and 6th symphonies were supplied to you on one CD? The fact is, you now own two symphonies by Beethoven. The fact that they are performed by the same conductor and orchestra may be of passing interest, but the really important information is that you now own two entirely independent compositions that happen to have been delivered to you in the same digital ‘envelope’.

In other words, my preferred approach to tagging classical music is to make the composition more important than the performer or the ‘package’ in which that composition arrived in your house.

Once you do this, however, you run into a bit of a problem: Beethoven wrote a fifth symphony …but so did Mahler (and Vaughan Williams, and Mozart and Haydn and on and on). If ‘composition’ is king, how do you distinguish between two compositions that have the same name? The answer to that is fairly straightforward: the composition title is not king (it can’t be, as it is duplicated much too often to be of such prime importance), but the composer is. If your “data hierarchy” is composer -> composition, then Beethoven -> Symphony No. 5 is very easily distinguished from Mahler -> Symphony No. 5. Bearing in mind what I wrote earlier about the joys of using composers’ full names, though, I’d refine that a little: I’d prefer you to actually tag things as Ludwig van Beethoven -> Symphony No. 5 or Gustav Mahler -> Symphony No. 5 and so on. If “composer is king”, then giving the composer a proper name, containing first and last components (and middle names if the composer was prone to using them in his or her life) is an important detail to get right.

Now, if you simply got these three principles clear in your head:

  • Use natural word orders and proper grammar
  • Forget what’s on a particular CD, as the composition is the main thing we’re interested in
  • In a well-formulated classical music metadata hierarchy, composer is king, composition is next in line to the throne

…then I think we’d already be able to tag our classical music collections much better than some have managed to do hitherto!

2.0 Tagging Specifics

With those three basic principles in mind, let’s consider what metadata tags we can provide and what metadata tags we should provide when cataloguing classical music. Because at this point, we are confronted with an uncomfortable truth: digital music tags were invented by people who had not the slightest idea about how things work in the classical music world! Let’s dabble in a bit of metadata history for a moment.

The first tagging scheme was invented to allow metadata to be added to MP3 (lossily-compressed) music files. Originally, MP3s had no way of storing metadata at all, but in 1996 Eric Kemp hit on the idea of tacking a small piece of unplayable data to the end of an MP3 file. To keep things small (remember, terabyte hard disks didn’t really exist back in the mid 1990s!), Eric kept his metadata chunk to a fixed 128 bytes. That had to do duty for storing everything you wanted to describe about that music -and in a very fixed manner, too. You were allowed just 30 bytes each for the title, artist, album, and a “comment”. And that really was it: forget about wanting to say someone was a ‘composer’, for example: the concept of a composer simply wasn’t thought about. So, right from the get-go, classical music listeners were forced to put their composer information into something labelled ‘artist’ …and since you were only given 30 characters to do it in, you’d expect to have to abbreviate heavily in a lot of cases. Thirty characters is plenty for someone like “Benjamin Britten”, for example… but is not such a good idea if you wanted your ‘artist’ to be Jean-Joseph Cassanéa de Mondonville… which is about 5 characters too long to fit! You were additionally allowed four bytes for a ‘year’ field and one byte to identify a genre -selecting from a predefined list of 80 values.

Incidentally, that list of 80 predefined genres is visible at Wikipedia and illustrates what we’re up against as classical music lovers! Of the 80 in the original specification, there is just one that applies to classical music… and it’s “classical”! The other 79 let you nicely subdivide ‘pop’ music into such wondrous genres as “trailer”, “acid punk”, “acid jazz”, “metal” and “space”, but get any thoughts of wanting to distinguish sonatas from symphonies from concertos out of your head! All three must just be described as “classical”!

Clearly, this first version of ID3 tags (as they are known) was rather too inflexible for its own good -and so, soon enough, a revised version of it was developed. ID3v2 implemented a variable-sized chunk of metadata at the start of a music file. Importantly, the total tag size was bumped up to a pretty massive 256MB, which essentially allows you the freedom to write whatever free text you fancy in any of the fields, of whatever length you like. Lots of new types of information could now be added to your metadata, too: each piece of information gets written into what’s called an ID3 “frame” -and there are over 80 frames to choose from, including such wonders as “Beats per Minute”, “Band”, “Lyricist” and so on.

Note that ID3 tags apply only to MP3 files. These are always lossily compressed. That is, chunks of the audio signal which are deemed too high or low or soft for a human to hear are simply removed from the signal and thrown away. What is left can then be compressed, in a manner similar to zipping up a Word document. MP3s are therefore usually pretty small (which is convenient) but they are not a faithful rendering of the original audio as recorded in the concert hall or recording studio. For a losslessly compressed audio format, you can instead store your music in FLAC files (other lossless compression formats are available, but I’ll concentrate on FLACs here, because they are open source, patent un-encumbered and non-proprietary, which makes them a safe bet for long-term archival use). FLAC files do not use ID3 tags to describe their data, but instead use what are called Vorbis Comments.

Vorbis Comments are much more flexible than ID3 tags. They consist of name=value pairs -and you can have up to 4 billion tags added to a single FLAC file, if you really wanted to! But within that restriction, you can invent whatever name/value pairs you like. If you want “composition_name=symphony no. 5”, that’s fine. So is “composer_name=benjamin britten”. Or “key_signature=E flat major”. If you really wanted to, you could even do “recording_engineer=John Culshaw” or “tea_lady=Mrs. Hudson”. 🙂 That said, just because you can add whatever weird name/values pairs you fancy, it doesn’t mean it is useful to do so: most music players out there will only display a fairly restricted subset of possible tags. If you add something exotic, it will be there in the file, and recoverable with the right tools -but your music playing software is likely not going to show it and will not usually be able to make use of it either.

So, to sum up at this point: no matter whether you store your classical music files in lossless or lossy formats, you are pretty much able to tag up your music files with an assortment of weird and wonderful tags -but many of the ones that are now regarded as ‘standard’ are only so because non-classical music listeners standardised on their use in the dark days of the 1990s. This means that there are an awful lot of “standard” tags that have absolutely no relevance to classical music lovers -and these should therefore not be used. Other tags are useful and need to be used -but have names which are liable to mislead or misdirect (requiring us to call a composition an ‘album’, a composer an ‘artist’ and a movement a ‘song’, for example).

I shall take this screenshot as an example of what most tagging programs expose as the ‘standard’ tags and explain for each one in turn how I think you should enter data into them:

(This screenshot comes from a program called ‘Easytag’, which is widely available on the Linux operating system -but it is very similar to MP3Tag, which is a broadly-equivalent Windows program. What it exposes here are really the bog-standard tags which you can expect any media player to interpret correctly and make sense of). Taking them in turn, then…

2.1 Title

This is the tag which describes the contents of this specific music file and no other. If you have four files making up a symphony, for example, whilst all four will have the same ‘album’ tag, each will have a unique title track, probably describing the movements of the symphony (such as “Allegro” or “Menuetto: Allegretto” and so on). It’s important that, in keeping with the idea that ‘using natural word orders and proper grammar’, you Don’t Introduce Unnecessary Capital Lettets Into Your Track Titles! This is true for any tag, really, but it’s very annoying to see -for example- a symphony movement described as ‘Allegro Con Moto’: Italians don’t all talk in Initial Capitals and Neither Should You. The correct form is, of course, Allegro con moto.

2.2 Artist

The Artist tag is a little problematic for classical music. What is an artist in the classical music environment? Is it Jacqueline du Pré, whilst playing a cello concerto? Is it Karajan when conducting a symphony? Or is the ‘artist’ the person we truly care about when we hear a piece of music on the radio and say ‘That sounds like Rachmaninov’ or ‘I bet that’s by Grieg!’ (Few classical listeners, I think, really listen to Radio 3 and idly muse “I wonder if that’s Mehta conducting…”!)

In other words, whilst it’s possible to fill this tag with details of the performers of the recording, I think it should be reserved for storing details of the guy or gal who wrote what is being performed. In classical music catalogues, in short, the artist tag should display the name of the composer of the work. I’ll have more to say about the specifics of what you put into this tag when I discuss the Composer tag below.

There’s a very good practical reason for sticking the composer name in the Artist tag, too: almost all music players will use the contents of the Artist tag to the exclusion of every other tag to determine the default sort-order of your music and how to display it. If you were to stick the performers names in here for classical music tracks, for example, you’d end up with Karajan’s recording of Beethoven’s fifth listed entirely separate (and a long way from) Zubin Mehta’s recording of the same work in most music players’ default display. With some players, you can fiddle around and get them to display and sort by the actual composer tag contents, but it’s never entirely straightforward -and as it’s not the default, there is always the danger of reverting to the artist display order at inconvenient times.

The short version, therefore, is that for classical music listeners, the composer needs to go in the artist field, no matter how strange that sounds. It’s not ideal, but that’s what the state of music player software really mandates.

2.3 Album Artist

This tag is also problematic! My source for understanding this tag is the article available here. And in case that link ever dies, here’s the relevant description:

‘Album Artist’ tags are stored inside mp3s and other music files. They denote the artist for a musical release, as distinct from artists for the tracks that constitute a release. That sounds like a technicality, so when do Album Artist tags become useful? Consider compilation albums. Each track in a compilation album may have a different artist, but the album as a whole should be viewed as a ‘Various Artists’ album. Therefore, we can tag the album with an ‘Album Artist’ tag of “Various Artists”.

 Now, I suppose this sort of thing can happen in classical music. Consider this CD in my collection, for example:

Now, each of the tracks on that CD are (on the whole) by different composers. If we are to use the ‘Artist’ tag to describe the composer, as I’ve suggested we should, in this case maybe we have to declare the composer as ‘Various artists’ and maybe the ‘Album Artist’ tag could then be used to describe ‘Wells Cathedral Choir’? Generalising from that: if we say Artist=’Beethoven’, could Album Artist=’Karajan and the Berlin Phil’?

Well, it could. But the problem with this tag is that different music players interpret it differently. If both Artist and Album Artist tag are filled in for a track in the way I’ve just described, it’s pretty much pot-luck whether your music player will display the ‘song’ as being by Beethoven or Karajan… and that won’t do at all! Here, for example, is what Foobar2000 makes of a track tagged in both the Artist and Album Artist fields:

…which is perfectly sensible. You can see on the right (one line up from the bottom) that the track has had an Album Artist specified, but the main library display on the left is not confused by this at all and correctly attributes the work to Beethoven.

But take a look at what Winamp does with the exact-same file:

It’s only claiming it to be a work by Karajan. No mention of Ludwig van is made anywhere else on that screen. And Winamp is not alone in this sort of behaviour. Here’s Windows Media Player:

Even though I’ve clicked ‘Artist’ in the left-most pane, it’s displaying ‘Herbert von Karajan’ as the Album Artist (because he is in this deliberately awful example of bad tagging). Fundamentally, what will be displayed when you fill in both the Artist and Album Artist tags is down to chance and the particular way your music playing software has been coded -and those are not things good data should depend on! So here’s my fundamental rule of tagging classical music: never, never add *ANYTHING* to the Album Artist tag field. It simply has no relevance to we classical music listeners -and we already have suitable places to describe the composer and the performing artists (as I’ll go on to describe below!)

2.4 Album

As far as classical music fans are concerned, this is a poorly-named tag, but it’s displayed by every single music player on the planet as ‘the name of the collection of tracks you’re about to play’. In classical music terms, therefore, since we play ‘compositions’, the Album tag should contain the ordinary name of the composition, with a consistent addition that I’ll get to in a moment.

By ‘ordinary name’ I mean that you’d say it was ‘Symphony No. 1‘, not ‘Symphony No. 1 in C sharp minor in 3:4 time and with a slight rallentando beginning at bar 8‘! Don’t clutter your album names with techie information that will seldom be of practical use: catalogues, even digital ones, are there to service queries you make of it. If key or time signature are things you genuinely want to query your music collection by (“Today, I feel it’s an A minor day”), then there is an argument to include that data in the Album tag, so you can see it and access it quickly. But bear in mind even so that you’re allowed 4 billion tags in a FLAC file, so an even better approach that keeps the Album tag clear of distracting information would be to create a ‘custom tag’ especially for “key=A minor” and “timesig=3,4”. When placed in their own tags, you should still be able to search by those criteria, even if you can’t see the information displayed in most players.

But in any case, most listeners are very unlikely to need to query by key or time signature, so I’d suggest just leaving that all out as mere clutter.

The one addition to the ‘ordinary name’ which I hinted at earlier, however, is an important one: most classical music listeners of my experience will have multiple recordings of the same composition, conducted or sung or performed by different artists. Thus we might have Beethoven’s Symphony No. 9 recorded by Herbert von Karajan and the Berlin Philharmonic and another recording of the same symphony by John Eliot Gardiner and the Orchestre Révolutionaire et Romantique. Since they are both called ‘Symphony No. 9’, how can your catalogue distinguish between them?

Well, it’s not really that difficult: you include the name of the distinguishing artist in brackets after the work’s ordinary name. Thus we’d have Symphony No. 9 (Karajan) and Symphony No. 9 (Gardiner). Keep the distinguishing artist short and sweet, just sufficient to make the two recordings distinct.

The distinguishing artist will usually be the conductor of the work in question -but many chamber works won’t have a conductor at all. In that case, the name of the ensemble playing the work might be used (thus, Schubert’s Octet might be tagged as Octet (Wiener) and Octet (Mullova). Sometimes, you will want to highlight the fact that a particular singer is involved: Callas fans might well want to know Traviata (Callas) as opposed to Traviata (Sutherland), for example.

Sadly, conductors sometimes decide to record the same work multiple times. Adrian Boult famously did two series of the Vaughan Williams symphonies, for example. So at that point, the conductor’s name won’t be enough to make things distinct: Symphony No. 9 (Boult) is not distinguishable from Symphony No. 9 (Boult), after all! Exceptionally, therefore, when the distinguishing artist name is not distinguishing enough, add the year of the recording (even the most perfectionist conductor is unlikely to record the same work twice in one year!) In this specific example, therefore, you might end up with Symphony No. 9 (Boult – 1958) and Symphony No. 9 (Boult -1969).

If you’ve any experience of working with databases, you will probably have noticed that the entire discussion about these tags is really a question of what constitutes the ‘primary key’ of a piece of classical music. What combination of attributes uniquely defines this piece of music as opposed to that one? My contention is that the true primary key for classical music is simply: composer -> composition -> distinguishing artist …with a possibility of needing to add a date to that last element in exceptional cases.

2.5 CD Number

Remember that the CD is to the music as Aunt Maud’s envelope is to her birthday card? The physical medium that your music arrives on is utterly irrelevant. Therefore the CD tag is utterly irrelevant too! It’s usually used to indicate ‘this is cd 1 of a 4-cd set’, ‘this is cd 2 of a 4-cd set’ and so on. But it’s irrelevant and should never be filled in! The fact that my copy of Götterdämmerung came supplied on 4 CDs is of no significance at all: when I rip the music to digital files, I’ll be ripping it as a single work into a folder that will end up containing lots of tracks, but none of those tracks will then be identifiably part of a set of CDs! You may sometimes see this tag also called the ‘Disc number’ or ‘Disk number’ tag… again, no matter what it’s called, it’s irrelevant, for in the world of digital music, we don’t deal with physical disks.

2.6 Year

This tag should be used to store the earliest date, where it’s discernible, at which the work was recorded. There’s no need to get very specific about the date: we do not need to record ‘Wednesday 15th December 1972′ for example; nor ’15-12-72’. Just ‘1972’ will do. It is unlikely that you will ever need the information you store in this tag: I don’t really know many people who sit down and say, ‘I fancy listening to recordings from 1963 today’, for example! And since hardly anyone queries the catalogue by date, it’s not really terribly important to be accurate about it. I have been known, for example, to just guess a date when I can’t find one mentioned in the accompanying CD booklet!

I’ll just add one word of warning, though: some music players will, by default, order their album lists by recording date, rather than by album name. This is, in my view, utterly a perverse thing to do at the best of times -and I would strongly recommend you get a new media player if yours does this (looking at you, Google Music!). For non-classical music listeners, being able to sort a group’s recordings by their release date is, apparently, a ‘thing’ -but it’s rarely if ever relevant to classical music listeners. Just be aware, therefore, that what you put into this tag might affect your playback experience.

2.7 Track Number

The track number tag should be considered compulsory. That is, everything you rip should be given a track number. Players will not order tracks correctly if they lack a proper track number, so you’ll find symphonies get played in Adagio – Allegro – Lento – Presto order, no matter what the composer intended, because the alphabetic sorting of the track titles will take precedence over absent track numbers!

Additionally, I’ll assert that the first track of an album should always start at track number 1. This seems obvious, but if it’s so obvious, how does this happen?:

Cello Suite No. 5 there starts at track 7… and that’s just wrong. The Prelude is the first track for that composition and needs to be given track number 1. How did it happen? Simply because (as you can probably guess from the track numbers associated with the Partita underneath it), both those works came supplied on a single CD, and there were 12 tracks on that CD, sequentially numbered, without regard to the change of ‘composition’ or ‘album’ half-way through.

This is not surprising: most CD rippers, left to their own devices, will rip entire CDs in one sitting, sequentially numbering tracks 1 upwards. Which is fine for the first work on a CD, but not so great for the second and subsequent works. To fix this, you need to learn how to rip your CDs in multiple passes, resetting the starting track number to 1 as each new ripping session begins. Either that, or you need to learn how to use a tagging program which allows you to re-number tracks starting at 1 after the entire CD rip is complete. dbPoweramp allows you to do this, for example; so do my very own Classical CD Ripper and Classical CD Tagger programs (both of which are free to download and use, by the way!)

And one final point I’ll make about track numbers: you should pad them to two-digits where possible. Whilst humans are very good at seeing the numbers 1,2,3,4,5,6,7,8,9.10,11,12 … 18, 19, 20, 21… and sorting them into ascending numeric order, computers can get this badly wrong: you’ll sometimes see things ordered as 1,10,11,12…19,2,20,21… 3,30,31 and so on (i.e., grouping all the ‘leading 1s’ together, then all the ‘leading 2s’ and so on). To avoid this, I’d suggest as a general rule padding your track numbers to two digits (hence 01, 02 and so on, rather than just 1,2). On occasion, I’ve even had to pad to three digits: an opera with a particularly large number of tracks finished at track 106. Therefore, all my track numbers for that work needed to be padded to three digits to ensure correct sorting: 001, 002 and so on. This is not common, however… perhaps fortunately.

2.8 Genre

This tag is meant to describe the type or flavour of the music file in question. We’ve already seen that in ID3v1, there was only one genre of classical music allowed: Classical! But this ignored the rich variety of ‘classical music’, which covers everything from folksong to lieder, song cycles, concertos, symphonies, large orchestral, chamber works, choral works, oratorios, masses and so on. Fortunately, these days, you are free to type all those sorts of things into the Genre tag.

Functionally, this isn’t terribly useful -in the sense, again, that I doubt many people sit down and ask themselves, ‘What concerto shall I play today?’, thus triggering a need to search their catalogue by genre. But practically, it can be very helpful to chop up a large body of compositions by a composer into separate genres where the genre defines the sub-directory on disk where the music files are stored. In other words, J. S. Bach’s output was huge and if you dumped all of it into a single directory called ‘J. S. Bach’, it would be playable and searchable just fine (digital computers work well like that!), but if you looked at it in a file browser, you’d be overwhelmed. With everything lumped together, you’d be very hard-pressed to be able to find a particular work visually, by mere inspection. But if we were to store Bach’s cantatas in a separate directory from his keyboard works and from his oratorios and from his orchestral works, then things quickly become more manageable.

For this reason, I use genre as the principle physical subdivision of a composer’s work:

That’s my file browser showing the directory structure for the composer Hummel: he didn’t write operas (or at least, I don’t own any by him!), so there isn’t a folder for that genre. Bach would similarly lack an ‘opera’ folder, but would gain an ‘oratorio’ one. Britten’s directory structure would include one for ‘film score’, too. The point is, whatever composer you’re dealing with, create as many genre sub-folders as are needed to break the composer’s work up into manageable chunks, rather than have everything piled in to one amorphous lump of files. (There are good practical reasons for doing this too, by the way: computers and file systems start slowing down dramatically when a folder contains a lot of small files. By distributing a given quantity of music files across half a dozen or more sub-directories, you prevent that slow-down from happening).

Having said that, there’s no particular science about what data can go in the genre tag. It’s up to you what you consider to be a ‘useful subdivision of works’ and what becomes a pointless display of pedantry! I mean, for example, that symphonies can be considered orchestral works: so do you have a separate genre for ‘symphonic’ and ‘orchestral’? Or do you just tag everything as ‘orchestral’. Similarly, are you happy with ‘opera’ as a tag? Or do you feel the need to sub-divide into ‘opera seria’ and ‘opera buffa’? I’ve written another short article where I list out what I consider to be useful genres: the biggest takeaway from that article is, I think, that:

…the genre provides a broad classification, not an academically-rigorous description of a work’s musical form. So, getting too specific, precise and (dare I say) accurate is not the goal when classifying a piece of music by its genre. You want something that helpfully chops a composer’s output up into maybe half a dozen to a dozen ‘style groups’.


The comments tag is where I would suggest you store the names and details of the performers of a work. It accepts free-form text, of course, but I would suggest you structure your comments text so that they always (or nearly always) are in the form of conductor, orchestra, choir, soloists, in that order. For example:

By the structure of that comment field, I know Roland Bader is doing the conducting, without me having to explicitly state the fact. I can also see the orchestra and choir involved, and I know all the individual soloists by name and voice type, too. Obviously, sometimes there won’t be all those elements present. A string quartet, for example, can just say something like ‘Gabrielli String Quartet’ in the comments tag -though I might also choose to name the individual players and their instruments.

Doing things in this way makes the players and their conductor invisible to most music players: they tend not to expose ‘comments’ in their usual interfaces -though that last screenshot is from the Clementine player, which shows you can reveal it there if you work hard enough and click all the right buttons in the right order! That’s true for most music players in use today, too. Remember, though, you’re dealing with computers now: they can find data, even when it’s not obviously displayed in a player’s interface. So, for example, I bet I can find Max Reger’s Requiem appear if I search for anything in which ‘Kawahara’ has sung:

Of course I can! It turns out she sang in three different Reger works, all from the same CD (no surprises there, then!). Her name isn’t visible anywhere in the program display, but the search engine has found her anyway.

2.10 Composer

In an ideal world, classical music lovers would use this tag all by itself to determine who wrote the particular piece of music in question. This tag would therefore contain correct, formal composer’s names and we’d not duplicate that information anywhere else. Unfortunately, as I’ve already mentioned, nearly all media players available today use the Artist tag to determine and display who a piece of music is by. Leave Artist blank but fill in the Composer tag correctly and, most likely, your media player will list the music as being by ‘Unknown artist’. Infuriating!

Now, technically, this practical fact means that the advice ought to be to fill in Artist and leave the Composer tag empty -on the very good grounds that it’s bad form, as far as data processing and management goes, to have the same information stored in two separate places. If you therefore choose to leave the composer tag empty, I won’t argue with you about it!

But many media players do let you display the composer tag, even if it’s not used by default to organise the player’s play list display. Therefore, I take the slightly less rigorous, but more pragmatic, approach of tagging both the artist and the composer tags with the same information. It is the only example of deliberate duplication of data in multiple tags you’ll ever see me propose as being ‘standard practice’. Both tags will therefore record the exact same data; both will tell us whether a piece of music is by Beethoven or Mozart or Bach.

I want to add at this point, however, that what people think of as ‘composer names’ has frankly astonished me in the past. You will see serious advice, seriously offered, by sincere people, to tag Artist (and hence Composer) as, for example, BRITTEN, Benjamin (1913-76). Or “Bach, JS“. Or “Mozart“. So I want to be very clear about this: composers have natural names. Use them in full. Use them with normal capitalisation. But leave out material that’s best read in biographies.

By this, I mean that the correct tag for works by Benjamin Britten is “Benjamin Britten”. By Bach, it’s “Johann Sebastian Bach”. By Mozart, it’s “Wolfgang Amadeus Mozart”. Natural names, naturally spelled and using natural word orders!

Natural word order is important because there’s absolutely no need to mangle word order in order to get a computer to find works by Britten, or to sort them all together. A computer will find ‘Britten’ whether you’ve entered it as “E.B.Britten” or “Britten, Benjamin” or “BRITTEN,Benjamin” or “Britten, Edward Benjamin” or even as “Baron Britten of Aldeburgh”. There’s thus simply no need in the digital age to use the reversed word order beloved of librarians and old-fashioned card indexes, nor to introduce capital letters where none are ordinarily required.

I strongly recommend you use a composer’s full name as ordinarily and commonly used by the composer himself. Thus, technically, it’s Franz Joseph Haydn -but neither Haydn himself, nor his family, nor his friends ever used the ‘Franz’ bit: it was simply a custom of the time to give a baby two saints names as first names and then to forget one of them. So it’s Joseph Haydn to you and your musical collection, just as it was to Papa Haydn himself. For the same reason, it’s not Edward Benjamin Britten. And although Mozart commonly called himself ‘Wolfgang Amadé Mozart’, the ‘Amadeus’ usage is now so commonplace that it would be churlish not to use it. Whatever you do, don’t, for heaven’s sake, start using his actual baptismal names: Johannes Chrysostomus Wolfgangus Theophilus Mozart is a mouthful too large to swallow!

For the same reason, don’t use a composer’s titles or honorifics. Vaughan Williams was immensely proud of having obtained a doctorate in music and accordingly frequently liked to be addressed (and signed letters) as “Dr. Vaughan Williams”… but don’t go calling him that in your music library. Edward Elgar, for the same reasons, loses his knighthood. Titles and honorifics go, really, for two reasons: they disrupt the expected sort order (you expect to find Elgar’s music amongst the other ‘E’ entries, not the ‘S’ it would be filed under if he was tagged as ‘Sir Edward Elgar’. Secondly, most people simply don’t think of them when they bring a composer to mind. People think of ‘Vaughan Williams’ or ‘Ralph Vaughan Williams’ or ‘Ralph’… seldom do they think of ‘Doctor’, let alone of the ‘O.M.’ honorific he was entitled to use after 1935. I suppose that’s the third reason for missing them out to: often, a composer only got to use a title or honorific for part of his life. Poor Ben Britten was a Lord (or Baron, if you prefer) for the last six months of his life, after June 1976. Tagging stuff he composed in 1936 with ‘Baron Britten of Aldeburgh’ is anachronistic (and pretentious!)

I will grant that there a couple of exceptions to this general ‘no titles’ rule: Lord Berners gets catalogued as Lord Berners, because no-one really knows him as anything else. His real name was Gerald Hugh Tyrwhitt-Wilson, but not only is that a mouthful, no-one will have the slightest clue who you’re talking about if you use it. So Lord Berners gets to keep his title. Sir John Blackwood McEwan also retains his knighthood, because the title is always and invariably used when discussing him or his works (Britten’s baronetcy isn’t, in contrast).

To try to help standardise on what I consider to be ‘correct usage’ of composer names, I’ve prepared a list of names in another article, which I recommend you follow at all times. It’s not exhaustive (that is, it doesn’t claim to list every single composer who has ever lived), but it is aiming to be prescriptive (that is, if you want to tag music by Janacek, for example, it aims to prescribe the fact that you should tag it as Leoš Janáček, complete with all the appropriate diacritic marks).

Following this advice will, of course, mean that your Mozart music will now be filed under W; your Prokofiev will end up under ‘S’ and your Beethoven will be filed under ‘L’. This is undeniably true and some people seem to hate this ordering. I can only suggest you persist: it’s actually good practice to get to know your composers well, and getting to know them by their first names seems the least you can do in that regard! Remember, too, that the computer is your friend: whilst your Mozart might be listed under ‘W’, you can always find it using your media player’s search capabilities and typing in just ‘Mozart’.

Remember, too, that everything I’ve just said about composer names applies to the Artist tag too.

2.11 Original Artist

This is another tag which non-classical music users will enjoy using and which classical music listeners should just leave blank, as it has no relevance to their artform at all. It is apparently common in the non-classical world for one artist to ‘cover’ the work of another. Not that I can claim to be expert in this, but as an example, Jimi Hendrix apparently re-performed Bob Dylan’s ‘All along the Watchtower’. In that case, Hendrix would be the ‘artist’ and Dylan would be the ‘original artist’. But there is simply no equivalent of this in the classical world: classical musicians spend their entire professional lives performing the work of dead people and it’s never considered a ‘cover version’ when they do! Leave this one blank therefore.

2.12 Copyright

I can’t imagine what utility there is to filling in data for this tag, so I simply wouldn’t bother.

2.13 URL

Similarly pointless, I think, so leave it blank

2.14 Encoded by

I don’t think you’ll have much meaningful data to enter here, but you’ll find that your ripping or tagging program will often automatically fill it in with various bits of self-promoting information (such as ‘Ripped by dbPowerAMP’, as a made-up example). Ignore this tag too, in other words,

3.0 Album Art

In addition to those ‘standard tags’ I’ve just enumerated, Easytag and other tagging software will almost always let you add Album Art to your music. It’s really important that you store your album art embedded within the music file and at as good an image quality as you can manage. Let me elaborate on that a little.

Windows users will probably be familiar with the concept of “folder.jpg“. Stick a picture of that name in a folder full of FLAC files and, left to its own devices, a lot of music playing software will use that JPG image as the ‘album art’ for those tracks. So, just source a good image of your CD cover art, call it ‘folder.jpg’ and stick it into the relevant folder, right?!

Wrong!! The trouble with that approach is that the album art is then a separate and distinct piece of data from the music files it purports to describe. It can be deleted or corrupted at any time -and then suddenly your music files won’t have any album art to display when they’re played. Worse, many music players (Windows Media Player, I’m looking at you!) will automatically run off to the nether regions of the Internet and fetch new album art that it thinks is most appropriate -and, at that point, it will generally over-write the folder.jpg file you carefully saved there, without a word of warning.

Album art is actually a really important piece of data to help describe your music: it will usually contain a lot of textual information that describes your music in quite technical terms. For example, whilst I might tag ‘Symphony No. 8 (Boult)’, the album art can tell me so much more:

Now I know it’s in D minor and on the Classics for Pleasure label -things I wouldn’t have mentioned in any of my textual tagging. It’s not often I need to know that sort of thing, but having correct album art allows me to know it on the off-chance I decide I do. So to have my carefully-curated collection of album art over-written by an over-mighty bit of software (or accidentally deleted by a clumsy bit of typing on my part!) is not a good thing to have happen.

Therefore: always embed your album art. That is, store it inside the music files to which it relates, so that it becomes an integral part of the music file itself. Doing this prevents accidents and malicious over-writing. It means that when you copy or convert your music, the album art will intrinsically go along for the ride. It means a directory listing of your music will look a bit like this:

Notice how all the tracks of ‘Missa Solemnis’ clearly have some sort of graphic associated with them -but there’s no separate ‘folder.jpg’ or any other independent graphics file stored on the hard disk in that folder. The art is inside the flac files, not a separate, external entity. Naturally, the album art is there displayed way too small to be of much use to anyone other than a particularly myopic ant… but you can usually right-click a file and open it in various tools and use that tool to expose the album art in its full glory, if you ever need to do so.

This approach brings the benefits I mentioned: the art will not suddenly be corrupted, deleted or over-written and it will always automatically go along for the ride whenever you copy or move or convert your music. The one drawback is that there’s now a graphics file embedded within every single FLAC file. Whereas before you could have had 100 FLACs and a single folder.jpg to ‘describe’ them, now you will have 100 copies of the album art stored within each of those 100 FLAC files. As a result, your music collection will take up more space than before. On the other hand, whilst this is undeniably true, the effect can be exaggerated. My music collection consists of around 56,000 files. A 1024×1024 album art picture is probably around 1MB in size, tops (in fact I’m exaggerating a bit, but let’s run with that figure for now). So if I embed a 1MB file in every single music track I own, I’ll be adding 56,000 x 1MB to my disk consumption… which is around 56GB. That’s non-trivial, but when disk space can be bought by the multi-Terabyte load at around £30 per terabyte, it’s really not a big deal in the scheme of things. In my case, I’d be adding 56GB to a music collection that’s about 1.2TB in size… an extra 4% or so, which is nothing.

One final parting shot: because album art is so useful, always try to acquire it at a large a size and in as good quality as possible. I always like to start with images that are at least 1000px x 1000px, but 500×500 is acceptable. Anything smaller than that, however, tends to end up being too small to be really useful. If that means you need to buy a cheap digital photo scanner, and manually scan the front cover of your CD jewel case art insert… go for it. It’s a very worthwhile thing to get really good album art.

4.0 What’s an Album?

I’ve already mentioned that the Album tab should be used to store ‘the name of the recording you’re about to play’, which is therefore, in classical music terms, the ‘composition’. But you do get edge cases where you have to decide what “counts” as a composition. If Mozart doodled 3-bars of harpsichord one day as a 5-year old, does that really count as a “composition” in its own right? I’d suggest not: it’s simply not significant enough in its own right to ‘count’, though it might well appear on a CD of his juvenilia as one 20 second track amongst a hundred others. Similarly, when Britten writes dozens of short folksongs, and Peter Pears sings them on a single CD, are we talking about 21 separate compositions, or not?

There’s no hard-and-fast way of answering that: you have to be flexible in your approach. Remember that the album tag is there to ‘group’ or aggregate together a bunch of separate tracks. If you think there is some standalone interest in a track or a subset of tracks, that or they should probably be classed as their own ‘album’. Conversely, if you never listen to a track except for the fact that it’s part of some larger collection of tracks, then it doesn’t have ‘independent existence’ and thus shouldn’t be classed as its own album, no matter that the composer may well have written and published it as a standalone work.

Take this “album” as an example:

It’s a collection of Britten folksongs. They were written over a period of several decades and there’s actually nothing to tie them together as an ‘integral work of art’, except for the fact that on this occasion Peter Pears is singing them all, accompanied by Britten at the piano. So, ‘The bonny Earl of Moray’ is definitely a composition in its own right (dating to about 1940, according to the Britten Thematic Catalogue). But does it count as an album in its own right, or not?

It’s not, in my view. Look at the lengths of those songs: the longest is only 4 minutes long, the shortest is just over a minute long. Those are not ‘worthy’ of being regarded as major compositions in their own right. Put another way, they don’t have the substance to be regarded as ‘standalone albums’.

When a collection of tracks becomes an album is thus not as clear-cut as you might like it to be, but let common-sense prevail. If you ever approach your music collection and say, ‘I fancy listening to X right now’, then ‘X’ is probably a candidate for being an ‘album’ in its own right. But if you only ever play a track because it’s part of something else, then the track stays a track and it’s the “something else” that gets regarded as the album. In the above case, for example, I would rarely think, “I want to listen to ‘The Minstrel Boy'”, but I have been known, from time to time, to say, “I fancy listening to some Britten folksongs”. So in this case, I have an album called “Folksongs (Pears)”, comprising 21 tracks, of which ‘The Minstrel Boy’ is just one. Separate compositions they may technically be, but they just don’t have the substance or heft to make them something I’m interested in, in their own right. Thus they cannot be separate albums.

I think that’s a bit of an edge case, though. For the most part, the dictum that “composition=album, movements=tracks” will stand you in good stead. But always use your commonsense and know that if you think of something as part of an aggregate, it’s the aggregate that deserves to be thought of as the album. And as a very rough-and-ready rule of thumb, if it lasts less than 4 or 5 minutes, consider whether it really can be thought of as an ‘album’, no matter that it’s a standalone composition. It might do, but if its really short, it probably shouldn’t.

5.0 Compilation Albums

Another edge-case I want to mention here:

How do you deal with this CD? It’s got 18 tracks, with each track being by a separate composer. Do you really have to split this up and rip it as 18 separate albums? Well, there’s a case to be made for doing so, at least in part: the Morten Lauridsen is an 8-minute long work that could definitely be ripped separately and treated as a standalone composition/album. It is, after all, a quite substantial composition in its own right. The Berlioz, Biebl and Briggs could also be given the ‘standalone album’ status, for similar reasons. But if you ripped those three as separate albums, would they lose something by then not being catalogued as part of the entire set of 18 tracks?

You must make a call on that on a case-by-case basis, I fear! In the specific case of the above CD, I decided that it would not make sense to rip any one work from the others. The point of the CD is, as a whole, to act as a ‘Christmas mood CD’, and all the tracks contribute to that mood, so I ripped it as a single CD. But what composer/artist did I then say was responsible for the works on the CD?

A lot of classical music rippers at this point tend to say, ‘well, it must be Wells Cathedral Choir that is the artist’. But it would be a mistake to do that, I think, because if you tagged it up that way, you’d now have Wells Cathedral Choir reported and displaying as a composer. You’re mixing what data goes into which tags -and that means you stop being able to query and view your data consistently. It would be as bad as storing ‘2016’ in the track number tag: you’ve now got a recording date in the wrong place, which is annoying in and of itself -but doing that compromises functionality. What if you then said, ‘query my music collection for recordings made in 2016’? Well, with the date stored in the wrong place, that music wouldn’t get captured by that search. So the same principle applies: if you use composer tag=artist tag, and you say that both tags will only contain composer names, you simply must not store non-composer data in those tags.

My principle answer to this conundrum is to avoid acquiring ‘mood CDs’, containing dozens of short tracks by multiple composers, which only make sense when taken in the aggregate, in the first place! But it’s obviously impossible to do so completely successfully -so my other answer is to allow a single exception to ‘only store real composer names in the composer and artist tags’: you are allowed to declare that you have music by that well-known composer Compilation:

It’s the only time you’ll see a non-real-name amongst my list of composers, and I dislike having to make any exceptions at all. But this ‘composer’ becomes the ‘dumping ground’ for a whole pile of ‘mood’ or ‘genre’ CDs which otherwise would be tricky to rip appropriately. For example:

Los Ministriles is a really good CD, consisting of a lot of short tracks of Spanish renaissance wind-band music. None of the tracks last very long, so they’re all under my ‘5 minute’ guideline for being ripped separately. Even if some of them passed that duration, though, ripping any of the tracks as separate albums to the rest would result in composers such as Juan Arañés being added to my list of composers -and he’s someone I’ve never really heard of and would likely never find another piece of music by again. You don’t want your list of composers being cluttered up by people who only have one or two tracks attributed to them -so again, it becomes a matter of remembering why we catalogue in the first place. Am I ever going to approach my music catalogue and say, ‘find me pieces by Juan Arañés’? Or am I more likely to say ‘find me that CD containing Spanish renaissance music’? If you are a fan of Arañés, be my guest and rip his tracks separately from the rest; if like me, you’ve never really heard of him, chances are, you’ll be better off ripping everything as belonging to ‘Compilation’.

Note, however, that when I am forced to use Compilation in the album tag, I do try to mention the actual composer (if it’s known) in the track title. That’s just in case I ever do get a sudden urge to search my music for stuff by Arañes, I can do a successful search anyway, because the correct metadata is present in the file somewhere. You can see this same approach (though unfortunately slightly different syntax!) in my tagging of the tracks on that Wells Cathedral Choir CD I mentioned before, too:

So I can still find all music by Rutter if I really wanted to, though it wouldn’t always be listed under the ‘John Rutter’ composer/artist tag. Search engines are good like that!

6.0 Foreign Language Issues

I want to mention at this point that many of the composers whose works you will catalogue won’t be English: shocking, I realise!! There will be plenty of Italians, French and Germans and likely a good smattering of Russians too (and my apologies in advance to the Czechs, Hungarians and every other nationality I’ve just lightly skated over!) This means that you will be cataloguing works such as Götterdämmerung and tagging tracks with titles such as Nell cor più non mi sento. Note the presence of diacritic marks over certain letters in both cases. These are not optional!

Miss off accents and other diacritics from certain words and they will become meaningless at best -or mean the wrong thing at worst.

There’s no reason to miss them off, either. For the most part, they are easily typed on any English or US keyboard, simply by using the Compose key functionality available in any Linux distro or by knowing how to type the relevant Alt+unicode number on Windows (for example, Alt+0252 will give you a u-umlaut). If you don’t know your Windows key combos, you should learn them. If you haven’t configured your Linux Compose key, you need to learn how to do so (though the Linux ecosystem varies so much that the article I’ve just linked to will only have specific relevance to few). And if you are a Mac user, you’re on your own!

In extremis, it’s potentially acceptable to use an English translation of the foreign-language titles. You could get away with tagging Wagner’s opera as ‘Twilight of the Gods’, for example… though I would suggest this is less than optimal, unless the performance is actually in the English language too. If I had a copy of a Reginald Goodall recording of that opera, then, I think ‘Twilight of the Gods’ might be acceptable. I wouldn’t use that name for Solti’s recording, though.

You can go too far in the foreign language game, though. Although I can read Russian quite well, most people can’t. And even when they can read it, when they’re playing music is probably neither the time nor place… so be prepared to transliterate. Yes, track 20 of Rostropovich’s recording of Lady Macbeth of Mtsensk should perhaps technically be tagged as “Создан полицейский был во время оно”, but I think you’ll generally find that “Sozdan politseysky byl vo vremya ono” is a tad easier on the (non-Russian) eye and more helpful to the on-Russian speaker/reader, too!

And one final point I want to make about the use of non-standard characters in tagging: never, ever use the “/” or “\” characters in your tag fields. The fundamental reason for doing this is that those characters mean something to the file system on which your music files are stored and you risk nasty things happening when you use a symbol for your own purposes that your operating system thinks means something else. A more subtle reason for not using them is that you tend to see it done when people are trying to squish multiple hierarchies of data into one field: for example, you’ll see people claiming that a piece’s genre is Classical/Symphonic/Nineteenth Century. At that point, they’re really trying to say three things about a work in one place -and that’s never good data management practice. The correct genre to pick at that point is actually just ‘Symphonic’: the ‘Classical’ is completely wrong since the Classical period of music composition ended in about 1800, and the Nineteenth Century is “derived information”: we know it to be 19th Century without you telling us, because you elsewhere tell us that it was written by Berlioz (for example). Derived information doesn’t need to be spelled out: it can, after all, be inferred from the other information you make available in other tags.

7.0 Summing Up

The above is a lot of reading, so I’d like to share with you an ‘executive summary’, that captures the gist of what’s just been discussed (and possibly adds one or two details whilst it’s at it!). I call them the ‘Axioms of Classical CD Tagging’, because I think they are self-evidently the best way of achieving a consistent, coherent and highly functional classical music catalogue. You are free to disagree, of course, but I’d appreciate you doing so with specific examples, so that I can learn from you more than the mere fact of your disagreement! Comments are accordingly open!

Axiom 1: The physical CD is irrelevant

The physical CD is simply the mechanism by which music files get delivered to you. It rarely describes the ‘work’ in which you are actually interested: one CD may contain a couple of different symphonies, for example; some Wagner operas will span 4 different CDs. So the physical medium on which music is supplied is thus irrelevant and that means the “Disc number” or “CD number” tag is equally irrelevant: never fill it in. You have digital music files now, not disks.

Axiom 2: Classical music’s primary key is: composer -> composition -> performers

The concept of a ‘primary key’ is simply a technical way of describing ‘what bits of information do I need to uniquely identify a piece of music I’m listening to, distinct from every other I might go on to play?’ The composer name is self-evidently insufficient: Mozart wrote over 600 works, so just saying ‘Mozart’ tells me nothing about which specific work of his you’re talking about. The composition name by itself is also not enough to be unique: both Mahler and Beethoven wrote a “Symphony No. 9”, for starters. The composer’s name in conjunction with the composition name is therefore needed to tell them apart. But then I can have multiple recordings of Beethoven’s 9th, conducted by different people. So to get to a unique recording of Beethoven’s 9th, I will need to specify the conductor or other principle artist (and sometimes, but rarely, a year).

In practice, this means that the COMPOSER tag should be the same as the ARTIST tag and the ALBUM tag needs to be the composition name plus a keyword description of the distinguishing artist (see Axiom 5). So, for example, Ludwig van Beethoven as composer and artist; Symphony No. 9 (Karajan) as the album.

If Karajan kept recording the same symphonies multiple times, however (he did!), then you may need Symphony No. 9 (Karajan – 1979)… but the use of the year in the album tag is only to be used when absolutely essential (otherwise you violate Axiom 7!)

Axiom 3: Composer names should be accurate and full

Since the composer name is a part of classical music’s ‘primary key’, it cannot be ambiguous. Label something as by ‘Bach’, and there’s a world of trouble ahead of you, as there were about 36 different Bachs who all wrote music at one time or another. So you need to spell out Johann Sebastian from Carl Philipp Emanuel… which means using full first names. And if you’re going to do it for one or two composers, you need to be consistent about it and use full names for all of them.

Tagging this way means your music starts getting collected and sorted in what might at first seem odd places: Britten will still be listed in the B’s (because Benjamin will do that for him), but Mozart will now appear in the W’s and Wagner will appear amongst the R’s. I will accept that this is quite a step-change in the way people expect their music to be sorted, but I recommend it strongly. There’s a certain charm in getting to know your composers by first name: my relationship with Mozart got a lot better once I started searching for him and thinking of him as “Wolfgang”.

Axiom 4: The composition is the Album (usually)

For the vast majority of classical music recordings, what counts as the composition will be obvious. It’s ‘this symphony’ or ‘that concerto’ or ‘that opera’. The names of those things then become the starting contents of the album tag. Note that Axiom 1 means a single CD may contain multiple “albums” (or may be just a part of a larger “album” that ships on 2 or more CDs).

A small minority of classical music is, however, often supplied in a form that would make it very complex to regard each separate composition as truly separate ‘albums’. A CD containing 23 folksongs, composed over the course of several decades, is unlikely to benefit from being ripped as 23 separate ‘albums’. In such cases, it is probably acceptable to rip the entire CD to a single ‘album’, attributed to one composer where that’s true, or to ‘Compilation’ where multiple composers are involved. There is no hard-and-fast rule about when this approach should be taken, but I generally regard tracks that last less than around 4 or 5 minutes probably not worth declaring to be their own albums.

Axiom 5: Recordings of the same composition will need distinguishing

To distinguish between the same composition being recorded by performed by different conductors, orchestras or singers, your album tag needs to contain a “distinguishing artist”, in brackets, at the end of the album name. Thus, your albums will be called things like “Turandot (Callas)” or “Turandot (Sutherland)”, or Symphony No. 9 (Gardiner) and Symphony No. 9 (Karajan). Usually the name of the conductor will work as the distinguishing artist. For concertos, though, it might be the solo violinist or cellist or flautist. For operas, it might be one of the principle vocalists. Even when you don’t have a need for distinguishing artist now (because you own only one copy of Turandot, say) you must take account of the fact that the need may arise later. So always include the distinguishing artist in the Album tag, in brackets, after the ‘proper’ name of the work. In rare cases, you may need to add a year inside the brackets to make multiple recordings by the same distinguishing artist truly unique.

Axiom 6: Tag grammatically

We Do Not Write In Initcap And Track Title Tags Should Not Use It Either. Be musically grammar-aware too -by which I mean, it’s not Finale allegro, but Finale: Allegro. You’re stating that this track is the finale of a work and it’s tempo is Allegro. “Finale allegro” just doesn’t mean anything musically!

Axiom 7: Don’t include one level of tag data in another

It is bad form (meaning: it can lead to bad outcomes) to repeat information you’ve included in one tag in a different tag. For example, If you’ve declared Artist=Bartok, and then discover you should have spelled it Bartók (with an accent over the ‘o’), then knowing you have to update the Artist tag is do-able. But if you’ve repeated that composer’s name in other tags, which perhaps themselves contain information in addition to the composer’s name, then you have a much harder update to perform, if you even remember you’ve got to update those other tags in the first place. In other words, it is easy to bulk-replace ‘Bartok’ with ‘Bartók’ if that’s the entire contents of a tag, but it’s much more complex to do a bulk update that replaces the word ‘Bartok’ with ‘Bartók’ when it’s part of a tag such as “Bartok: Four Romanian Folk Songs”.

For similar reasons, it is wrong to include the composition name in every track title. So it’s “Allegro con brio”, not “Symphony No. 5: Allegro con brio” -because you’ve already declared that it’s Symphony No. 5 in the Album tag. And don’t tag the third movement of something as “3. Andante” or ‘III. Andante’: we already know it’s track 3 because the Track Number tag tells us (or ought to, if you’re ripping your music properly! See Axiom 14), so it doesn’t need to be re-stated in the Track Title tag.

Axiom 8: Don’t specify the obvious

A Flute Concerto, by virtue of it being a Flute Concerto, is obviously for flute. And by virtue of it being a Concerto, it’s equally obviously going to be for orchestra. So tagging it as “Concerto for flute and orchestra” is pointlessly redundant. “Flute Concerto” tells us everything we need to know in three less words. The main reason, indeed, for not getting to wordy (apart from it being slightly pretentious) is that whilst screen real estate is abdundant on a PC or laptop monitor, on many playback devices, screen real estate is very constrained and thus space is at a premium, to be used carefully. Consider one of those vacuum flourescent displays that show the track title, and scroll it from left to right, for example: you’ll be waiting for Christmas before the word ‘Flute’ comes up if you’ve chosen to mention the ‘Concerto for’ bit first. On a smart phone, things can be even worse:

You tell me what is the primary instrument in the first two concertos mentioned on the top line here? Whereas on the bottom line, even apart from the Album Art, you can readily tell that it’s a violin concerto. If brevity is the soul of wit, so it is also the heart of efficient tagging. So don’t state things which don’t need stating: it just takes up room!

Axiom 9: Word order is important

This one relates to Axiom 8, but isn’t necessarily very obvious. When you tag a concerto, make the principle instrument appear first in your album name: Flute Concerto or Violin Concerto, for example, rather than ‘Concerto for Flute’ or ‘Concerto for Violin’. The reason is simply that you want the important information (what principle instrument is involved) up-front in the album name… because you’ll sometimes be using play devices which don’t have enough room to display everything. On a smartphone, for example, you are likely to be width restricted. Tag things as ‘Violin Concerto’ and ‘Flute Concerto’ and even if the names get truncated to fit, you’ll likely still be able to see the words ‘Flute’ and ‘Violin’. Tag it the other way round, however, and you’ll likely see everything listed as just plain ‘Concerto fo…’

Axiom 10: Stick to musical facts

We are music listeners, not theatrical impresarios. I will allow that we might want to know where Act 1 of an opera finishes and Act 2 starts, but knowing that track 16 is “Act 1 Scene 5 part 1” helps my music listening experience not at all. Keep stage directions out of it, too. And we don’t need track titles to tell us if a chorus is singing or it’s a piece of recitativo –your ears should be good enough to spot the difference. Track titles should similarly not contain dramatis personae names: if you don’t know that it’s Dido singing at one point and Aeneas at another, follow along with a Libretto or Score until you do. For the same reasons, we tag things as being by ‘Benjamin Britten’, not “Benjamin Britten (1913 – 76)”, since his birth and death dates belong in a biography, not a composer/artist tag.

When you’re searching for or playing music, your database needs to be full of musical facts, not non-musical distractions.

Axiom 11: Genres should be judged carefully

Unless your name is Wolfgang Amadeus Mozart, you almost certainly didn’t write ‘classical music’. Tagging everything as having a Genre of “classical” therefore makes about as much sense as declaring that henceforth all oranges, apples and pears will just be known as ‘fruit’. So don’t over-simplify. Meanwhile, use Genre to split large collections into meaningful chunks, but not get so detailed that they split things into a bazillion tiny chunks. Use my list of genres as a good starting point.

Axiom 12: Non English-speakers used diacriticals. So should you. But don’t go overboard, either

Wagner didn’t write Gotterdammerung, but Götterdämmerüng. If you don’t yet know how to use your computer to type accented letters quickly, learn how to. Given all the Germans, French, Spanish and Italians who have been writing classical music all these years, it’s pretty essential. But don’t start tagging Russian music in Cyrillic or Japanese music in Kanji. That’s not likely to be useful… but it does look a bit pedantic and too-clever-for-its-own-good! (If you’re a native Russian or Japanese classical music listener, feel free to ignore this axiom!)

No matter what your nationality is, however, never use ‘/’ or ‘\’ in your tags. Their use is potentially dangerous (as those characters mean something to your operating system), and in any event usually indicates confusion about correct tagging practice.

Axiom 13: Album Art is important. Get it and embed it

The album art displayed when a media player is playing a track is important. Make it good quality and a decent size: 500×500 pixels is a minimum; 1000×1000 is better. Always embed your album art in the music file itself: do not use files such as ‘folder.jpg’ to act as your album art, because it will likely be over-written, corrupted, deleted or become separated from your music when you copy it or convert it.

Axiom 14: Every Album starts with a first track. The first track should be numbered “1”

The fact that it was supplied as track 16 on a multi-work CD is irrelevant (see Axiom 1). If you don’t know how to renumber tracks when ripping, or if your ripping software doesn’t let you do that, then learn how to use the auto-number function in your favourite tagging program and re-number the tracks post-rip. In either event, every single ‘album’ you possess should start at Track 1.

Axiom 15: Beware of Custom Tags

You will change media players sometime in your life. Your tagging strategy needs to ‘play to the basics’ so that it works (as far as you can reasonably foresee) with most of them. That means not creating a bunch of custom tags which 8 out of 10 players won’t read, won’t display correctly, or won’t let you search by anyway. If you are relying on ‘silent conversion’ of tags between players, so that what one player displays as “title” another will correctly translate to “track name”… well, don’t rely on things you have no control over. Software might get the conversion right today; there’s no guarantee it will continue doing so tomorrow. Stick to the ‘standard tags’, however, and you should be pretty safe whatever software player you end up using.

That said, if your music appreciation skills are very high-level, such that you care about things like key, time signatures and so on, by all means use custom tags to accommodate your exacting standards. But keep that sort of data cruft out of the standard tags that the rest of us care about and use daily!

Axiom 16: Physical storage needs to follow the tag strategy, not stay independent of it

I’ve seen suggestions in other tagging guides that if a CD contains two symphonies (say), then you mark them up as two albums, but rip everything into a single directory. So the tags logically split the one CD into two albums, but physically, it remains in one place. This is a bad idea. For a start, see Axiom 1: the physical CD is irrelevant.

But also, see Axiom 14: you’re supposed to label Symphony No. 1’s first “allegro” track as track 1. If you’re also supposed to label Symphony No. 2’s first allegro track as track 1, you’re now faced with the prospect of one directory on disk trying to contain two files called 01-Allegro.flac. Since that won’t be allowed by your file system, this physical storage strategy can’t be allowed. Only if you rip Symphony No. 1 to a ‘Symphony No. 1’ directory, and Symphony No. 2 to a ‘Symphony No. 2’ directory will you be able to have two track 1’s called Allegro that don’t conflict or threaten to over-write each other. So make the physical storage follow your logical tagging.

The complete ‘physical storage strategy’ should really mimic your tagging practice. Composer is your principle directory; split that up into multi genre-based sub-directories; and within each genre directory, create folders for each album, suitably distinguished from each other by ‘distinguishing artist’ names in brackets. Thus, you should aim for this sort of thing on your file systems: