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

Last.fm turns 20

theverge.com

Post reply on HN
Print view View original

151–160 of 178 posts

Re: Last.fm turns 20

#151

Post by vintermann »

kerblang wrote: post #102
ep103 wrote: post #61

Earlier quoted context omitted.

That's the way music-map.com works, and it is absolutely fantastic. But no, that is NOT the way these other services recommend things. Spotify, for example, recommends music that it thinks is related, but not necessarily liked by both groups of people. For example, a person who likes Black Sabbath, and really technical instrumental metal, probably doesn't like the comparatively simpler slipknot. While some people who…

> For example, a person who likes Black Sabbath, and really technical instrumental metal, probably doesn't like the comparatively simpler slipknot That's just thinly disguised metal snobbery. Almost all metalheads adore Black Sabbath, because they wrote fantastic songs . Sabbath's music caught on because it's so accessible. They were certainly good musicians, but by no means anywhere near the top of the Technical Sop…

Metal snobbery is a legitimate taste, though. It's the sort of thing a recommendation algorithm should consider.

Re: Last.fm turns 20

#152

Post by vintermann »

SimonPStevens wrote: post #51
AdmiralAsshat wrote: post #21

Still use last.fm. Most "You might like this" algorithms go straight for the low-hanging fruit, and often fail to take any kind of nuance into account. I can't tell you how many of the streaming services will see my Black Sabbath play history and immediately recommend, "If you like Black Sabbath, you should love...Slipknot!" But I've never had a real person make that mistake, because a real person who looks at my las…

I had always assumed that recommendation algorithms didn't work on genres, but instead did a lookup based on something like: "You like X" "Person Y also likes X" "Person Y also likes Z" "I should recommended Z to you" (proportional to the number of people there are who liked X and Z, compared to other options) If this is the case (and perhaps I'm wrong), it's nothing to do with the algorithm not understanding the sub…

> an average of all humans likes, which will never fit your exact personal unique preferences

Well, it's not as if that's an unknown problem in machine learning. In all generative domains (and I suppose recommendations are a generative domain in a sense) there's a risk of "mode seeking", where the model produces bland/blurry/gray output in order to cover its bases, and not miss too badly. It's rarely what we want.

There's also the often quoted "Most of the volume of a high-dimensional orange is in the peel, and the average of two vectors is very unlikely" (Well, I like to quote it because it's fun to say. I forget where it originally came from.) In plain English, if you need more than two dimensions to characterize variation, then you need to use spherical interpolation to find a likely data point in between them, not regular interpolation.

It's surprising how many times this has been overlooked and rediscovered.

Re: Last.fm turns 20

#153

Post by alex_sf »

dougSF70 wrote: post #100
ep103 wrote: post #61

Earlier quoted context omitted.

That's the way music-map.com works, and it is absolutely fantastic. But no, that is NOT the way these other services recommend things. Spotify, for example, recommends music that it thinks is related, but not necessarily liked by both groups of people. For example, a person who likes Black Sabbath, and really technical instrumental metal, probably doesn't like the comparatively simpler slipknot. While some people who…

Spotify recommends me tracks that I have already added to my playlists. It knows me really well.

I dunno, if I was writing the Spotify recommendation engine and wanted to trick people into thinking it was good, I'd do the same thing.

Re: Last.fm turns 20

#155

Post by lucifer13 »

ChrisArchitect wrote: post #69

Scrobbling is fun, just one of those 'personal data' things to look at from time to time. Last.fm was always corporate/sketchy tho and never got into it/didn't want an account on it. GNU Libre.fm worked fine for years tho! Nice to have an archive of data just sitting on there. (Lack of hands to work on it to get other features like dumps and more analysis never panned out tho). But since the pandemic my listening hab…

As mentioned elsewhere on this page, ListenBrainz project (https://listenbrainz.org) is an open source alternative to Last.fm. Its run by the MetaBrainz Foundation non-profit. You can also import your scrobbles from Last.fm and Libre.fm to it.

Re: Last.fm turns 20

#156

Post by TurboHaskal »

I used it for a while. The recommendation engine needed work, and I felt the social aspect of the whole thing was influencing the music I was listening to.

These days, I would go as far as to claim that I grew to despise the figure of the music nerd. I never visit websites such as last.fm and RYM, and the mere sight of a copy of "In the Court of the Crimson King" is enough to ruin my day.

It's hard to explain.

Re: Last.fm turns 20

#157

Post by CrypticShift »

kixiQu wrote: post #134
beezlebroxxxxxx wrote: post #38

Earlier quoted context omitted.

I've found most recommendation systems in general don't grasp the distinctions in genre that are in metal, or the stylistic differences. Most stuff is just tagged "METAL", when there is a huge difference between grind, funeral doom, or atmo-black. Even within a genre there can be differences it doesn't really capture, like tech-death: you might call Beyond Creation and Warp Chamber both tech-death, but in reality the…

How do you find the https://everynoise.com/ categories to fare at this?

This is Spotify’s list of "genres".

To be fair (and more exact) Spotify data scientists detect patterns in the audio spectral features and use them to manually create/add new genres. The "collective wisdom" nature of last.fm and RYM will often give better results/phraseology.

You could say that Spotify could also be data mining these Music Social Media sites! My own research does not show this to be true (at a large scale anyway). But they should (and pay them for it)

Re: Last.fm turns 20

#158

Post by CrypticShift »

sph wrote: post #149
Rodeoclash wrote: post #90

Earlier quoted context omitted.

Two things I like to see trailed in music recommendation algorithms would be: 1. Associations happen at a song level. At the moment when artists are recommended you get what I call the "Red Hot Chilli Peppers" effect. That is, the artist that basically everyone seems to have in common in their playlists is RHCP. Instead, I think the recommendations should be more granular, at a song level. Artists can diverge quite w…

> Weight follow on plays more highly What if I listen to Black Sabbath and then my niece wants to listen to the Peppa Pig soundtrack after that? What if it's a party and everyone is queueing their favourite song? The issue with recommendation engines is that they're dumb as rocks. They cannot tell that the Trentemoller remix might be related to War Pigs, but not Peppa Pig, without ears and a brain. So, to be safe, th…

What you are talking about (simple statistical aggregation) may be true 10 years ago. Deep learning is truly different.

I'm not saying it is "smart" or listen the way we are. But is not "dumb" either in that it does not just follow prescribed rules.

I personally prefer human curation. But I've heard many many stories of how spotify's AI sometimes surprises people with uncanny magical recommendations (and dumb ones too). And it is just a start.

Re: Last.fm turns 20

#159

Post by CrypticShift »

rchaud wrote: post #125

One of the first "Web 2.0" apps I remember using next to Gmail; it was called Audioscrobbler back then. I used the scrobble feature religiously to log my listening habits from 2005-2012. While Last.fm failed to capitalize on early innovations like the social media features and algorithmic online radio, the "scrobble" feature oddly enough stood the test of time. Even though it was the pre-smartphone age, developers bu…

> failed to capitalize on early innovations.

CBS really screwed this up. To their credit they kept the API alive, so other could somehow innovate with the data.

Re: Last.fm turns 20

#160

Post by CrypticShift »

gtirloni wrote: post #74

I don't understand why Spotify or some music streaming service doesn't buy Last.fm already. It's a pretty useful service that nobody is going to pay for unless it comes included in a streaming service already.

Because they don't need to: They chose machine intelligence (= The Echo Nest, ) over unreliable Human tagging (= Last.fm)

Also, Spotify now has the data of more than 400 million active listeners. Last.fm has at best 20 millions of users (including a lot of inactive one).

Post reply on HN