It's rare that a day passes in our house without music being played somewhere. But because we're now working on the launch of Syntonetic's Moodagent playlisting application for iPhone and iPod Touch, I've been spending even more time than usual listening to my own music, thinking about playlists, and what makes a good one.
In my previous post I wrote about the digital music dilemma: when you've accumulated so much digital music that you can't remember all the tracks in your library, how do you decide what to play? Picking tracks at random can sometimes be interesting, but it doesn't work when you want to hear a particular kind of music.
Of course, that description of the problem glosses over a vital issue. What defines a "particular kind" of music? For some situations -- like music to accompany an exercise workout, or to listen to while running -- tempo is crucial. To meet this need, Potion Factory’s Tangerine can create iTunes playlists based on tempo, based on the BPM (beats per minute) metadata field. But if you're not working out, dancing, or managing a restaurant, a playlist based on tempo alone may sound too monotonous.
If tempo is too limiting, is music genre a useful way to characterize different kinds of music? Now we have the opposite problem. Wikipedia needs four pages just for a list of music genres, and obscure genres are a legitimate target of satire.
Even a thoughtful discussion of genres is likely to generate more heat than light. But that discussion illustrates how digital music, and iTunes in particular, has focused people's attention on subject of music genres, and how to fix them. You can download scripts to help you remove unwanted genres from your tracks, and read articles that help you create your own custom genres. There are programs like TidySongs and TuneUp to help you clean up your iTunes metadata. In particular, the TuneUp software ...
... allows you to filter down the number of different genres that TuneUp tags for. There are currently 3 Genre Levels. Level 1 is the smallest with 20 Genres, Level 2 is 250 Genres, and Level 3 is the max with 764 Genres.
Reading that TuneUp Tip piqued my curiosity: how many of those 764 genres are present in my own library of 8500+ songs? This turns out to be the perfect project for iTunes Smart Playlists, so I made a Smart Playlist for each of my genres. I now know that, despite the vast numbers of genres that may exist in theory, 98.5% of my own music is classified into just 31 genres -- with 1.5% marked as unclassifiable, or having no genre assigned.
What's more, the genre distribution is highly skewed: all but 5% of my tracks are Rock (2745), Jazz (835), Blues (798), Pop (529), Folk (483), New Age (436), R&B (353), Soundtrack (342), Alternative & Punk (307), Country (253), World (236), Reggae (160), Classical (156), Books and Spoken (132), Latin (101), or Alternative Rock (100). The last 5% is divided (unevenly again) among 15 other genres, ranging from Easy Listening (80) all the way down to Indie (1).
Some of these genre classifications are ridiculously, even hilariously, incorrect, so I'm going to look at that aspect in a future post. But just the distribution of tracks among genres is enough to convince me that (as with tempo), I can't create many useful playlists on the basis of genre alone, except perhaps for some special occasions.
Sometimes I may want to listen to an hour or two hours of Jazz, Latin music, or Reggae -- we still miss Inka Inka, even though the last time we heard them perform was in 1996. But the problem is, I've got 8 hours of Latin music, 11 hours of Reggae, and three whole days of Jazz. And my rock music would make a playlist three or four times the length of Woodstock.
Clearly just picking a genre will not alone make a good playlist, even using the shuffle option to pick tracks at random (which may make sense to a computer, but not to a human listener). I either have to pick songs manually (which is limited by my ability to recall all the tracks in my library), or get some software that is smarter about music, and can suggest tracks that go well together.
So in my next post, I will make good on my previous promise to review three playlist generators that aim to pick tracks that go well together: Apple's Genius for iPod and iPhone, Syntonetic's Moodagent, and MusicIP Mixer from Amplified Music Services, another playlist generator that creates acoustic fingerprints.