iHeart Radio? The child of Clear Channel? The conglomerate that bought up all sorts of regional stations and homogenized them to the point of seemingly choking regional sounds and styles in the favor of slick, corporate approved campaigns to consolidate listeners into pens like cattle to the slaughter[1]?
My take, as a musician, is that this isn't about helping listeners find "Music they love" - if it does happen, a lucky accident for the listener. It's fundamentally trying to exploit algorithms and human nature to be able to charge more for ad space. The business hates risk, and this modelling and "learning" is simply to reduce their risk of playing a track that might encourage a person to, you know, change the station.
This may sound extremely cynical, but it's hard to deny that Clear Channel is a very spurious company playing the same old "We love music and musicians" card when they're really talking about their own profits.
From what I remember Clear Channel purchased a company called Thumbplay and rolled it into iHeart Radio to build out their web app. The developers that they acquired with that purchase whom I've met legitimately are music fans and had very mixed feelings about having to work for Clear Channel.
Just an anecdote, I totally agree with your sentiment.
Oh I can understand! I'd feel for their plight too; it took me a while to get out from a couple gigs that I had principles in conflict with, so "doing it for the paycheck" isn't a light subject. Thanks for sharing, especially as a reminder that where we start and why we start may not be where we end up in the long-run.
Agreed. They seem to have some pretty smart people behind the curtain, and I imagine the work is probably quite interesting on an abstract, technical level. But I've always found iHeartRadio to be one of the most obnoxious listening experiences imaginable every time I've been a captive audience of it.
I must be weird because I like music across a wide range of styles, electronica, rock, rap, r&b, pop, world, and all mixtures of them. I get bored being stuck in an algorithmic bubble with similar sounding music for too long. I like music that is "interesting, according to me" but I can't really define interesting until I listen to it.
I'm similar - I have an eclectic taste in music that is often not served well by static, long-term algorithms that place me into a "well" of the same sounding music. It's a problem I think about a lot.
Matrix factorization can deal with this a bit by using the high dimensional space to place your tastes into an area that reflects many different styles at the same time. In part 2 of the blog post we're going to talk about how we're modeling the acoustic qualities of music, which can find common patterns from completely different genres (for example, you may like soothing music with female vocals in both jazz and indie rock). In part 3 we'll talk a bit about how we can combine recent signals (like thumbs) to take into account your current mood, which I find helps pinpoint interesting music to surface right now.
> we're going to talk about how we're modeling the acoustic qualities of music, which can find common patterns from completely different genres
That's neat! I was curious of this is / was being looked into. It seems like I often get music that's matched based on a demographic (if that makes sense), rather than music matched on the characteristic features of the current song / band.
What we call genre is sometimes circumscribed by stylistic elements, sometimes by subculture (demographics as seen from the listeners' own perspective). But quite often since the rise of radio, it's circumscribed by target demographic as seen from advertisers' perspective.
The worst case of that is probably "new age", a label rejected by virtually all the artists so labeled (and most of the listeners), and having no common traits to speak of, but lumped together as whatever sold better in bookstores than in record stores.
>how we can combine recent signals (like thumbs) to take into account your current mood
This is the most interesting part of the problem to me. I always worry that the signals I send my favorite music radio service permanently alter the course of the channel. In some cases, I want that. In others I don't.
Imagine able to easily explore the entire music space, say with a set of sliders: "I want more vocal", "faster beat", "more bass", "edgy", and get a selection of songs that matches this.
They're mapping the multidimensional music spectrum. I highly doubt the end goal/use is to firmly place you in some 10d music coordinate system so you can listen there forever. You need the mapping before you can do anything interesting, like a cool auto generated real life soundtrack that uses data from your pulse, phone calls, paycheck, work hours, interactions, etc. :D
I think clustering in vector space works really well for many things, but not for discovery of new unexpected music.
The only thing in my experience that works for that is algorithms that take advantage of human curation (people with similar tastes). And even then filter bubble is a real thing (Facebook)
The problem is the granularity. Everybody wants to separate based on genre when what we really need is to separate based on song and possibly then reverse to artist.
After you downvote "Faithfully" enough times because it makes you gag, Journey leaves your list.
Except that "Frontiers" and "Edge of the Blade" sound nothing like the garbagey Journey ballad schmaltz that everybody sings on karaoke night. So, you will never hear those songs in spite of the fact that you may like them.
The problem is that you can't get this information without throwing the occasional curveball at a listener. And, from the streaming app point of view, it is very risky to throw new songs at a user that you don't know definitively that they like because they might dislike it enough to change the channel, app, etc.
Consequently, recommendation engines run by corporations will only ever be totally safe and boring.
> The problem is that you can't get this information without throwing the occasional curveball at a listener.
This is the problem with machine learning or most algorithmic recommendation schemes. There are no curveballs, no randomness. Of course you want one part of the experience to be similarity.
I really want the equivalent of visiting a friend's house and them putting on a mixtape and a couple of the tracks or artists suddenly jumping out at you. Or going to a gig and being blown away by a support band. Or a dinner conversation about a new band your friend just discovered, or ...
That probably won't ever happen - I can't see an algorithm recommending Julie London if you usually listen to Moby, but some of those bizarre leaps are often the best discoveries.
I definitely disagree about curveballs. Coming across stuff you don't like is part of discovering what you do like.
On your first trip to the record store, you won't have any idea what you're buying. It might be good, it might not be good. After a while, you get to know the owner of the store, and they learn your tastes, and you learn a bit more about music and start to know what will be good and what won't be good. Sometimes you'll both make a mistake and you'll still buy something you don't like. And you'll learn from that.
Why shouldn't our learning algorithms work the same way?
You should give My Favorites Radio a shot on iHeartRadio. Aggregates all the tracks you have thumbed up across all genres and throws in some new recommendations once in a while. Its basically a lean back personalized radio station just for you.
If someone used What.CD to generate new music it'd be better. That place is nuts about classifying music and sorting things. Best music site on the net.
You're not weird, and this modelling experiment's unsavory undercurrent is that it's trying to figure out how to keep you on iHeart Radio / Clear Channel stations...and nowhere else...for all your music listening needs. Basically, if it isn't on their system, you won't know it exists. Walled garden to the extreme.
Agreed, music map is a great interactive way to discover music. Last year one of our internal hack week teams built something like that using our vector space model. What was cool is you could switch between the different types from the UI, e.g. Nirvana -> Red Hot Chili Peppers -> DC101 Alternative Rock. You could also start stations from the UI - it was really fun discovering music with it!
Glad to hear you like it, because I think in some foreign countries (according to my accounting) their system is introducing my material at a respectable rate. They aren't getting 'pushed' like based on an ad campaign, spike-and-valley style. It's much more that one or two listens might happen - granted, this could be from other 'touch points' but going back in time before I was as coordinated account wise. Can't claim it as fact, but gut instinct I suppose.
My take, as a musician, is that this isn't about helping listeners find "Music they love" - if it does happen, a lucky accident for the listener. It's fundamentally trying to exploit algorithms and human nature to be able to charge more for ad space. The business hates risk, and this modelling and "learning" is simply to reduce their risk of playing a track that might encourage a person to, you know, change the station.
This may sound extremely cynical, but it's hard to deny that Clear Channel is a very spurious company playing the same old "We love music and musicians" card when they're really talking about their own profits.
[1] Just one article - it's been covered time and again. http://www.daveyd.com/articlesclearchannelbyjeffpearlstein.h...