MP3 Shuffle algorithms

[Jon.]

Does anyone know anything about these?

I have a 1GB MP3 player made by SanDisk, and I have about 420 songs on it. I like to listen to them on "shuffle" mode - kind of like having a radio station where I know I'll like all the music. The only problem is, about 70-80% of the time, it plays one of 20-30 songs that I keep hearing. And if I try to zap past it, it usually (about 80% of the time) plays the same song.

Anyone know if this is a defect that could be fixed or just something I have to live with?

Thanks!

 

[jmercer]

Something you'll probably have to live with, I'm afraid. Most MPG players have such algorithms encoded in firmware on the player, including the IPOD Shuffle... which does a better job of randomizing songs than other products, I have to admit.

 

[Dilb]

I've heard anecdotal stories about some products having algorithms that randomly select a folder with equal probability, then randomly selecting a song from that folder. The person had 1 song in it's own folder, so it came up as often as all 20 songs together that were in there own directory. So it might help to delete any folders that seperate your music. Or it might just mess up whatever internal order your have.

 

[Jon.]

Thanks, Dilb, but I'm just using one folder. I suspect there's a flaw in the firmware, and they just don't care enough to fix it. After all, it doesn't take much effort to zap through a few songs to get to one I haven't heard recently.

 

[Psi Baba]

I've been wondering about this too. I have a Dell DJ and its shuffle feature works well enough but does have odd quirks that I wonder about. For one thing, I've determined that it indeed selects an album first, then a song. I know this because it always starts with the first album as it would appear alphabetically, but not always the same song. At first, I didn't realize this because I was always getting the same song first, until I realized that song was a one-off, and after I deleted it, I was always getting a song from the next album (now the first) in the list.

Another thing is that I often get what I call "two-fers," or two consecutive songs by the same artist, although they may be from different albums. It seems to happen a lot. It could be just that I have a disproportionate number of songs by those artists.

So far, it doesn't appear to repeat tracks unless I restart the shuffled playlist. Can't say the same for Windows Media Player, however.

Does anyone here know what kind of algorithms these players use, or do they use look-up tables of some kind or what? I've heard that some may have a kind of learning mode where songs played more often have a higher chance of being selected, which could explain Jon's problem. The bad thing about that is it's an ever-tightnening loop: songs played more often get selected more often. If they are selected more often, they become songs that are played even more often yet and so on. I wonder how you could reset that.

Unrelated to shuffling, one feature I'd like to see added to mp3 players is the ability to create exclusion lists as well as playlists. I'm starting to add some spoken word and podcast files to my player and I don't want them popping up in a shuffle when I'm listening to music.

 

[Jon.]

Unrelated to shuffling, one feature I'd like to see added to mp3 players is the ability to create exclusion lists as well as playlists. I'm starting to add some spoken word and podcast files to my player and I don't want them popping up in a shuffle when I'm listening to music.

On my player, you can have the shuffle setting set to one folder only. You could create two folders in your root directory: one for "shuffle-able" tracks, and one for others. (This is what I've done, BTW, calling the first folder Radio Jon.)

 

[jmercer]

Keep in mind that - without special equipment - most random number generators utilize a "seed" value that comes from some source which is not random. You can do a lot with an algorithm to add to the "randomness" of the choice, but it's very difficult to make them truly random. The better the algorithm, though, the more the perception of true randomness.