Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Spotify’s Discover Weekly: How machine learning finds new music

hackernoon.com

Post reply on HN
Print view View original

241–250 of 275 posts

Re: Spotify’s Discover Weekly: How machine learning finds new music

#241

Post by travmatt »

anidesh1512 wrote: post #212
ar-nelson wrote: post #210

Earlier quoted context omitted.

These solve the problem somewhat, but, in my experience, they only play songs I've already listened to. So listening to songs outside the genres I want recommendations for will still ruin Discover Weekly's usefulness as a discovery system. I used to love Discover Weekly, and found a lot of great music through it, but it's gotten less useful over time. A few things that would improve it: • Multiple Discover playlists…

They intersperse some new songs between my already saved songs.

I prefer them for exactly this reason - I’ve learned that I actually prefer less novelty in my playlists than I previously assumed. Spotify seems to understand this, and incrementally alters my playlists, moving through tracks and doing a good job of removing tracks that I skip continuously. I’ve noticed these playlists actually do a good job at discovering music for me within each genre

Re: Spotify’s Discover Weekly: How machine learning finds new music

#242

Post by erikbern »

markussss wrote: post #140

The biggest problem with the Discover Weekly is its inability to understand _why_ you're listening to a specific subset of music. It might not be that your taste in music suddenly changed, or that you discovered a new genre that you're incredibly interested in, even though your most listened to genres or songs changed for a few weeks, or your listening patterns changed for a few weeks. A few examples: I'm Norwegian,…

We had to build some explicit filters for this at Spotify. For instance we blacklisted Christmas music from being part of the listening data – until them people would get a ton of Christmas music recommendations every January. But to your point, there's a long tail of other contextual things.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#243

Post by GreenPlastic »

My problem with Discover Weekly is it weights recent listens way too heavily. I was listening to R&B for awhile and really loved the new songs it was pulling up - then I switched to jpop and then the list was entirely jpop. Over time, if you don't listen to other music it becomes kind of eclectic / coffeeshop pop.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#244

Post by capkutay »

erikbern wrote: post #239

I built the foundation of this system while at Spotify. While it's true that we looked at a lot of different signal, at the point when I left (early 2015), it was all based on collaborative filtering. The reason collaborative filtering works so much better than anything else is that given enough data, it will already encompass everything else. If there are reasons why certain users prefer certain sounds, or certain l…

I guess this is a less technical question - but what would your advice be to an artist who wants fans of their more famous influences (similar artist) to discover their music? Aside from the obvious (tour, market yourself to popular playlists etc).

Sounds like there's some NLP and web-scraping involved...so would it make sense to come out with blog posts that compare you to the famous artists that influence you?

Re: Spotify’s Discover Weekly: How machine learning finds new music

#245

Post by ProfessorLayton »

chewz wrote: post #24

With all that AI excitement Spotify, Google Play Music and Apple Music misserably fail at generating good musical recomendations. At least for me. Spotify Discover Weekly was no different for me when I still used Spotify (six months ago). Not only it is not personalized at all it is also quite lame. I am doing enough to seed - following enough artists, genres, liking album, subscribing to playlists but the quality of…

At bare minimum I really with I could tell these services which artists I explicitly don't like so that they're never recommended to me. That alone would be a huge improvement.

Disliking multiple songs from the same artist only to have it suggest the next one from their repertoire is infuriating. Worse, Apple Music will play songs you've disliked in the radio anyway.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#246

Post by GreenPlastic »

nobleach wrote: post #172

I find that my music tastes can vary too much for Spotify. While working, I mostly listen to Electronic music. But I can also take a hard right turn and listen to Hall and Oates... or Fleetwood Mac, or Steely Dan. Then some form of 80's throwback. But as far as recommendations go, I really only want to expand in that first genre of Electronic Music. I don't want Jethro Tull to be recommended to me. My feed used to be…

I actually want the opposite - I don't want the stuff I listen to all the time to be recommended to me, I want some of the one off stuff I listen to to appear in the feed at around the same frequency.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#247

Post by wcarron »

mattkrause wrote: post #234
wcarron wrote: post #217

Earlier quoted context omitted.

I get repeat songs now and then. Not a lot, but a noticeable number.

Perhaps you've listened to the regular version of a song, but the Discover Playlist has surfaced its "radio edit" or some other very, very similar version?

Most definitely not. Same songs. Not referring to radio edits (though I have seen them too, sometimes).

Re: Spotify’s Discover Weekly: How machine learning finds new music

#248

Post by ExactActuation »

criley2 wrote: post #138

Machine learning, lol. Ok. I can write a Discover weekly in about two lines of psuedocode. IF user.history.containsEvenOne("rap/r&b") THEN discover.weekly.setAll("rap/r&b") TA-DA~ you now have the complete "Discover Weekly" experience.

Hahahahahahahahahahahahahahahahaahahahaha.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#249

Post by n42 »

scrooched_moose wrote: post #200
pj_mukh wrote: post #175

Earlier quoted context omitted.

I really wish I could get Spotify to "forget" or "reset" my preferences. When I first started using Spotify, I listened to a lot of slow calming music (because I was working). Now ALL my Discover weekly's are just 'Iron and Wine', 'The National' derivatives. Its made Discover Weekly useless to me.

I have the same issue with Iron and Wine and I don't get why. Every recommendation engine I've tried decides I absolutely love them, no matter what seed I start with or what songs I like/dislike. I don't mind Iron and Wine but they are definitely a 'meh' for me, so it gets frustrating when they dominate every playlist. And, disliking/thumbs down never seems to get rid of them. Are they generic enough that the algorit…

You might be on to something. Iron and Wine might be the Kevin Bacon of music recommendation algorithms.

Re: Spotify’s Discover Weekly: How machine learning finds new music

#250

Post by ricardobeat »

cr1895 wrote: post #56
ricardobeat wrote: post #23

Discover Weekly works well enough for me - I do wish I could exclude devices from influencing that playlist as what I listen while gaming on PS4 is completely different from the rest of the day. It also seems to have a bias towards “big” commercial releases. Spotify’s “intelligence” in general is a huge let down though. Radio stations are extremely limited - more like 15-song static playlists indefinitely on repeat!…

>It also seems to have a bias towards “big” commercial releases. I haven't noticed this bias at all. It might be due to our own music preferences.

I don't listen to pop very often - that's how I notice it. Eventually I'll realize that a song it added to my playlist is playing in the background in a shop, on TV or the radio.
Post reply on HN