Musical Preference Algorithms

0 Members and 1 Guest are viewing this topic. Read 1118 times.

Syrah

  • Full Member
  • Posts: 580
Musical Preference Algorithms
« on: 14 Jan 2016, 06:26 pm »
Does anyone know a good site where someone can fill in band they like and it predicts or recommends other band you might like?  The website equivalent of the nerdy college friend.  I think this is the reason so many people stop musical discovery after they've left school, you can keep your mind open and discover new music - but it's so much more work than it was.

I know some music services try to do this, but if they do it the way Amazon or Netflix do it - it's extremely simple and it's usually just done by genre.

It would be great if somehow we could analyse music on a deeper level and determine what it is exactly that makes someone love or hate something - then use that to guide them to musical nirvana.

It's particularly difficult for someone like me who is not entirely committed to any particular style or genre (although I lean in favor of jazz and indie rock) maybe there's something in a style I don't normally like (say country) that I might love.

Thoughts?

megabigeye

  • Jr. Member
  • Posts: 99
Re: Musical Preference Algorithms
« Reply #1 on: 14 Jan 2016, 08:40 pm »
I'm kind of in the same boat as you.  I used to have a lot of buddies that I could get recommendations from.  Most of my friends now, though, just don't have the same musical tastes as me.

I used to think Pandora did a decent job of introducing me to new music.  Now, though, it seems like it always just plays the few songs that I "like" or that I use as station seeds and eventually just defaults to playing the Beatles.  I can't tell if Pandora's gotten worse or I've just become less tolerant/patient.

Allmusic.com also has recommendations based on your likes and dislikes if you have an account.  I haven't played around much with it yet, but it seems okay from the little bit I've poked around.  This plus youtube might be a decent way to discover new music.

Also, record stores!  There is one guy at my local place that always has some recommendation when I check out.  "This is a great album!  Have you ever checked out _____?"  Either he and I have very similar tastes or he has an encyclopedic knowledge of music.  This is one of the main reasons I always go back.

I've also thought of starting a "Recommend Me..." thread here on AC.  Lots of music lovers here.  Seems like it would be a great resource.

mcgsxr

Re: Musical Preference Algorithms
« Reply #2 on: 14 Jan 2016, 09:13 pm »
The website I use is called Tastekid.  It does tunes and movies and likely more. 

I found it very useful when getting into sax jazz initially. 

Syrah

  • Full Member
  • Posts: 580
Re: Musical Preference Algorithms
« Reply #3 on: 20 Jan 2016, 01:03 pm »
Thanks guys.  That Tastekid seems pretty good.  It's not complex algorithms, but it seems to at least do a good job of hitting similar types of music.  I keyed in a few obscure ones, like Broken Social Scene, and it did manage to hit quite a few other bands I like.

Someone needs to come up with something more complex, for example if you chose bands of the same genre that you liked and disliked, it could dig deeper down and then cross genres.  It could even do that by comparisons with other people or by some kind of analysis of the music itself - i.e. people who like John Coltrane, the Clash, and Tchaikovsky (like me), usually like ___ .

Maybe one day we'll just mail in a DNA sample, and it will tell us what we like.

But thanks for the tips!

gnostalgick

Re: Musical Preference Algorithms
« Reply #4 on: 28 Jan 2016, 06:19 pm »
   Pandora has always been awful.  I remember putting on MC5 & The Stooges and getting Foghat :o, Echo & the Bunnymen plus Devotchka equals Starship 'We Built this City!' :duh:

   Tried Last.fm years ago & although they made no "One of these things is not like the others, one of these things just doesn't belong" mistakes, the track selection was pretty limited.

   Anyways, even though I'm currently subscribed to Tidal, Spotify's recommendations were good enough that I constantly considering going back (very well might if I get a more forgiving dac/amp/speakers one day).  I discovered many great ambient/electronic/industrial-soundscape/quiet-drone artists through them.  The issue is that after a month of listening to new sounds, they forgot I also liked indie pop/rock/etc.  Can't comment too much on jazz or classical, but its at least competent: listen to Charlie Parker, get Dizzy; listen to Ravel, get Debussy.

   Lately I've been using 8tracks.com which is all user-submitted playlists.  Admittedly they're rather hit or miss, but you can usually tell a bad one by three tracks in, and the good ones can be very good indeed.

   I've read many good things about Roon, but the price seems very steep considering everything that's available for free.

Kirk

ASCTLC

  • Jr. Member
  • Posts: 82
Re: Musical Preference Algorithms
« Reply #5 on: 28 Jan 2016, 09:45 pm »
Try Slacker.  I can't stand Pandora.  I didn't get but a few days out of Pandora before I got sick of it repeating the crap out of everything and going off on a music direction that isn't even close to what I stated I wanted to hear.  Slacker has been doing me great for years!  You're going to get repeats no matter what but it's a matter of how many.

From what I understand (I'm no expert on this) Pandora is computer predicted music list where Slacker is human based.  The human element knows better what fits and is related than a computer can.

The "What are you listening to?" thread has sent me off on a wonderful experience of new bands/music.

Andy