Thinking a little more about the soundtrack post from a couple of weeks ago, I was struck by the fact that I don't seem to have the same strong associations with more recent songs that I do with some older stuff. It's not that I'm buying less music, I don't think, but rather that iTunes and the lack of good radio has changed the way I listen to music.
In particular, I miss good radio, and I wonder if it would be possible to get iTunes to simulate the sort of thing I'm after (explained below the fold).
In my opinion, a really good radio station playlist breaks down sort of like this:
- 50% recent songs (last six months, say)
- 40% really good songs from the last several years
- 10% oddball stuff that the DJs like
The first category is part of what fixes songs in my mind as associated with a specific time and place. It's not just the super-mega-hit songs that are completely inescapable-- "Hey Ya" a few years ago, for example-- but the general repetition of whatever they're playing at the moment. "Hardest Way Possible," listed in my soundtrack post and known to almost no-one else got a fair bit of airplay from WEQX a while back, which is what got me to buy it.
The second category gives you a wider sort of context. You end up relating songs to each other, and seeing connections between new hits and good tunes from a few years back. The third category just provides a little individuality-- you get the occasional album track from a ten-year-old record thrown in, to spice things up.
The current problem is that I really don't have a radio station that does that. Well, a combination of that, and the fact that I'm not really satisfied with the current state of pop music categorization. WEQX is pretty good, but it only comes in well in my car, and I'm not 100% happy with the music they play. KEXP goes overboard on the "wide variety" thing. They're great for discovering new bands, provided you listen really closely, because they hardly repeat anything.
The other issue is that I can't really get iTunes to do what I want. I tend to shuffle-play (well, "Party Shuffle") my "recent purchases" playlist when I'm at home (unless I have a new purchase that I particularly want to hear straight through, like the Hold Steady record this past week), but this has the problem that songs purchased a month ago are dead to me, no matter how good they are. At work, I tend to shuffle-play the four-and-five-star playlist, but given that there are 2730 songs on that list, it doesn't bring up recent stuff all that often, or repeat tracks very quickly. And it doesn't do anything about the 4122 songs that still aren't rated.
What I'd really like to be able to do is construct a weighted playlist. I'd like to be able to set things up on iTunes so that if there's a probability p of any single track coming up, the probability of a song rated four or five stars is 5p, and the probability of a song added in the last six months is 10p (or something like that-- I'm pulling these weights out of the air).
I have absolutely no idea how to do that on iTunes, though. I can construct smart playlists for the different subsets of songs that I'm interested in, but if I merge them and shuffle-play the result, they all get equal weight, which isn't what I'm after.
Does anybody know a trick for making a weighted playlist in iTunes?
- Log in to post comments
Hmmm... For a start, you could change the "Genre" field in your metadata tags to reflect those categories. Instead of "Rock," "Indie," or whatever, you could change the genres of selected tunes to "Recent," "Classic," and "Oddball." Then, you could create a smart playlist that includes those three genres, set to a Maximum of, say, 50 songs, sort by Least Recently Played or Least Often Played, and enable Automatic Update.
This would give you a mix of those three artificial genres, that would automatically sort for things you've haven't heard in a while. That keeps the playlist fresh. Of course, you'd need to manually make sure that you tag those tunes in the proportions 50/40/10. You might get the occasional playlist that's mostly Oddball, but it should average out over time.
I believe there's also an iTunes setting that lets you choose different randomizing algorithms, based on the Album tag -- for example, you can set it so Shuffle doesn't "randomly" select six tunes in a row from the same album. I've never played with this, but maybe you could experiment with these settings along with modifying the Album field to get the effect you want.
Another trick I use is to use the "star" rankings as a kind of code for updates I want to make in iTunes. 1 star means I hate it and want to delete it from Library. 3 stars means I want to change the genre to something else. Etc. That way I fine-tune my mix.
Personally, my favorite mix is "100% oddball things that the DJ likes," but I've never used music as a way to keep up with the zeitgeist. I have a "radio"-style automatic playlist that also includes short spoken-word sound clips, comedy, some vintage advertising that makes me nostalgic, etc. That way, it's not just a constant wall of music, and I find I can listen comfortably for longer periods.
Smart Playlists allow you to limit your list to a certain number of songs, right? So what if you made a Smart Playlist that picked 50 songs from your New Stuff list, another SP that picked 30 songs from your Old Favorites list, and another that picked 20 songs from your Oddball list? And then a Master SP that picked songs from those three lists? Would that do the trick?
I think I'll try it myself...
Paging Skwid to the Smart-Playlist shaped phone. (Sorry, old Usenet habits.)
What Scott said. I've been doing this forever, and Joel is on the right basic track. Since my iPod is significantly smaller than my music library, and I listen to my music primarily in Shuffle mode, I want a mix that is similar to what you've just suggested, although our needs vary slightly. Since my limiter is space, rather than song number, I assemble my playlists based on size, but the principle is roughly the same, and laid out in the comment linked above. I can give more specific advice if I know how you've got your library constructed and whatnot...