Monday, March 10, 2008

Tag, Mozart, You're It!

I love my iPod. And iTunes. And my iPhone. The iPod literally changed the way I store and listen to music in a way that CDs never did. When I decided to retire at the start of 2002, I took a road trip down I-5 from Portland to San Diego to deliver and install an iMac for my early grade-school age nephew. Being on my own, I was able to listen to whatever I wanted to, and I took along a lot of CDs and talking books (most on cassette - remember them?). Were I to proportionalize that media today, it would be the size of a 1/2" LMG counter, back then it took up as much space as both Combat Commander boxes, perhaps a little more.

The problem I have had is that iTunes is really not oriented around multi-movement works, as is found commonly in classical music (although progressive rock has it's share). I guess they took their cue from early CD players, which often included an "index" number that was to subdivide tracks on a CD. Of course, back then CDs seemed to be a major improvement in sound quality (partially true), while the iPod has been until recently a step *back* from CDs in that the music is compressed.

That's interesting to me, and actually a lucky accident, as the one genre of music that suffers most from compression (at least to my abused ears) is classical music with it's wide dynamic range, which is to say the differential between the loudest and softest sounds. Most popular music rarely requires such a wide range, although to be fair the difference in compression schemes tends to focus on compression speed which relates to how quickly the music is recorded rather than the bit rate, which directly affects dynamic range. 16 bits seemed like a lot back in 1982, now most music is recorded using 24-bit technology (although there are few players that take advantage of it - SACD and DVD Audio have never caught on with the buying public other than audiophiles who listen to - wait for it - classical music).

Which is a long and convoluted way of saying that I've not ripped my sizable classical collection of CDs, but it's been because of time issues rather than sonic issues. Because of the improved compression schemes (I'm using 256kbs AAC formatting these days when I do rip a CD, which is very rare), this is the time to start. However, the root problem remains - CDDB, the user-generated CD database that allows you to bring in song data based on the usually unique track signature of a given CD, is horribly inconsistent in how it captures that data.

For example, let us say I am ripping a recording of Beethoven's setting of the Missa solemnis, which is a five movement work, and the discs I'm ripping it from also contains a Mozart Mass as well. The first four movements, located on one disc, might look like this:

Name: Beethoven; Missa solemnis in D, Op. 320 - I Kyrie
Artist: Robert Shaw; Atlanta Symphony & Chorus
Album: Beethoven; Missa solemnis in D, Op. 320 - Mozart; Mass in C minor, Op 36
Composer: Ludwig von Beethoven

The second CD, which contains the Agnus Dei movement, might look like this:

Name: 5. Agnus dei - Beethoven; Missa solemnis, Op. 320
Artist: Robert Shaw; Atlanta Symphony
Album: Mozart; Mass in C minor, Op 36
Composer: Beethoven, Ludwig v. (1775-1825)

I am typing from memory, so some of the Opus numbers and dates may be a little off, but this is for illustrative purposes so bear with me.

As you can tell, there's a lot of difference between the two discs. Worse, some of the fields (the first example's Album field) are so long as to be virtually useless on an iPod, where you only can see around 15 characters at a time.

Now picture this happening with every single album that is out there. Some will use Roman numerals for the movements, which often are named after the tempi they are in, and thus produce an interesting alphanumeric sort, some include the composer's name (iTunes does have a Composer field, but it's useless for popular music and I'd rather not have to remember that I need to think differently when browsing classical material), so data is repeated ad infinitum. In other words, the usefulness of the CDDB is largely negated when you're ripping classical material. Worse, you have to come up with a naming convention that will work in most situations for consistency. When you have several hundred classical CDs, as I do (advanced music degrees will do that to your collection), you want to be consistent from the start or do the work over and over and over.

As the choir I am in is performing the Missa solemnis along with Beethoven's "Choral Fantasy" Fantasia in C (sounds dirty, is actually pretty dull for the choir but a great piano work), I wanted to have these works on my iPod so I can listen to them in the car. That means I have to rip them, which means I'm thinking about a consistent naming convention. I've come up with this, please comment as I suspect I'm getting close to a ripping run in the near future.

Name - Will contain *only* the movement number and title. In single movement works, this will repeat the title of the work itself (e.g.; "Toccata") without any catalog data. I will include key information if appropriate.

Composer - Last name, First name (DOB-DOD). This one is pretty easy, and having the lifespan included is nice to have if you're trying to win a bar bet.

Album - This is the biggest departure: it will contain the *work* rather than the physical *album*. The format will include the last name of the composer (with initials if there are multiple composers with the same last name, such as JS Bach and his progeny). Format is Composer: Work. I will include catalog information in this field. This allows me to scan quickly for a given work, and even though there are multiple recordings of several works in my collection, in general I can differentiate by including that information after the work if necessary. I rarely have multiple recordings by the same artist/ensemble, if any.

Artist - Possibly the most useless information, although I know a lot of classical music lovers consider this the most important. Once I've gotten a good recording/interpretation of a given work, I really don't care who has done it unless someone wants to know who it is. In many cases, the recording label is a better differentiator (Naxos, for example, a pretty good line of budget CDs that I highly recommend). Oddly, this is the best recommendation for ignoring the album a recording is from, as often a different set of artists will be involved with one work and not another (especially vocalists, or if an additional recording is a soloist rather than an ensemble).

The fly in the ointment is, of course, what to do when you have a compilation of works on an album. These are usually "vanity" productions to showcase an ensemble or artist, sometimes from less-well-known organizations. Technically, the Missa solemnis recording mentioned above is one of these, as it includes another work, even if that work is also "major". In general, if the works are multi-movement, they will generally be separated out (which requires me to edit the number of discs as well as the track numbers as well as all of the other data). In the case of particularly well organized discs such as one I own that has Vaughn Williams' "Fantasy on a theme of Thomas Tallis" as well as Barbur's"Adagio for Strings", I still have to decide if I wish to break up each work individually or not. I guess I could keep such a group together as a playlist, although I prefer to save those for albums I've recently purchased so that they don't get lost in the shuffle (no pun intended). Because I buy something like four to eight albums a month, and it's easy to forget you've bought something in the 60 Gb of data sitting on my iPod.

Anyway, every album becomes a lot of work, and even though I suspect most people will not be terribly interested in classical music who read this blog, it is an interesting database problem. If I've learned one thing about gamers, we like puzzles and we are technically minded, and this is exactly the sort of thing that I would (and am) taking to gamers to get their take on.

That, and I figure if nothing else I'll drive off anyone who's still reading after my diatribe on evangelical Christianity, at least the flavor that my daughter has latched onto (no new information on that front, btw).

If you've made it this far, feel free to add a comment if you think that you can help me out here. I really want to start getting this music into the database, as I rarely listen to classical anymore because it's in a format I rarely use. Thanks!

1 comment:

Eric said...

First, classical music is a lot of work. It's not supposed to be easy. Why do you think audiophiles listen to classical? Because they've put a lot of work into training their ears to hear things nobody else can. So they have to show their talents off by listening to something hard.

So, suck it up and enter in that data by hand, dammit. It's what an audiophile would do.

(Seriously, though, there's got to be 57 different ways to categorize classical music, and it's likely different across publishers and composers. If you're trying to normalize that data the chances of it being done the same way for any two random releases in a free-for-all style database like CDDB have got to be close to nil.)

Now as far as remembering the new stuff you've bought, just set up three or four smart playlists keyed off the date the music was added to iTunes. Something like:

This month's music
Last month's music
Two month's ago music

Every time you sync, you'll have the last few albums you've bought in those playlists.

(Now, that DOES bring up the one thing that drives me crazy about playlists on iPods. Why can't we browse playlists by album?)