Every Noise at Once – an algorithmically-generated scatter-plot of musical genre | Hacker News Reader

Every Noise at Once – an algorithmically-generated scatter-plot of musical genre

everynoise.com

71 pointsnyodeneD23 comments

This is awesome. There is also this old thing: http://techno.org/electronic-music-guide/

Similar effort, but purely editorial and only about electronic music, with commentary

Thank you! I've been looking for this since the first time I found it I don't even remember how long ago! I found it really interesting.

With so many categories of rock, you have to be careful about how you angle your guitar pick when hitting the next chord; you just might derail into another genre by accident.

This is amazing! I wonder how it would compare to the algorithm that Spotify uses to suggest music to users. I've always found Spotify's recommendations to be very tightly banded and think they could benefit from more genre-hopping.

The snippet for Italian punk is neither Italian nor punk, kinda makes you wonder about the quality of the original data. Probably it would make more sense to coarse grain a little bit more if you can't trust the data.

Hah. There WAS an Italian punk band called Arturo, but that "Arturo" definitely wasn't them. Fixed. This is a perpetual source of trouble, but finer distinctions actually help deal with this, as they produce more-specific audio profiles, which gives the filters a better chance to recognize a mis-attributed song as an outlier for the genre...

This is awesome, didn't realize you can actually click to get a sample which is amazing. Most stuff I checked out was correctly filed, too.

Would be cool to ad a crowd sourcing element to suggest changes to get an even better base.

An undo for the "listened to this" icon would also be helpful.

Searching for my favorite band (Savatage) they are in roughly the right categories (all sorts of metal) but in sum too far away from classical. Dunno how to express this better, seems like a complex problem though.

This is nitpicking, but it's kind of weird that there are tons of genres I've never heard of but no modal jazz.

Very cool! However, the "zydeco" sample wasn't typical zydeco at all. I guess it was Wayne Toups, and I'm sure he can do some real groovy zydeco, but this wasn't. ;-)

How 'glitch' came to be placed so close to 'classic Dutch pop' escapes me... Fascinating to hunt around in though! Much more fun than writing my dissertation!

Probably a lot of that is the particular choice of projection into 2D space.

This is an old project, but nevertheless really good. The 'related' artists are spot-on, and it has lots of obscure bands.

Not entirely relevant, but I was quite pleased with the example tracks for Black Metal and Doom Metal.

More than that, It's very interesting to look at one of the (numerous) metal subgenre and see how the various band are mapped, defining tendencies in the subgenre. It works very well in power and doom metal, in particular.

It puts "breakcore"[1] between "boy band" and "new romantic". I do see some similar genres in the area (deep breakcore, intelligent dance music), but I reckon a dimension has been squashed which was essential in positioning the entire area of music around IDM. They lie scattered between genres of music with which they have little to no connection.

[1] https://www.youtube.com/watch?v=JF7qmAMkksA

The blurb doesn't seem to imply that genres close together are in any way related; just that they have some similar characteristics. It doesn't strike me as entirely improbable that there are some surprises there.

That's right. This map begins as a scatter-plot, not a force-network, so particularly in the middle of the map individual pairwise distances in these two dimensions can be perfect or misleading. The readability adjustments to keep the names from landing on top of each other also distort the data in a technical sense.

But if you go into a genre's own page and scroll down, there is a little inset map of that genre's immediate neighborhood. That view is restricted to the most similar genres according to ALL the dimensions, so breakcore's neighborhood, for example (http://everynoise.com/engenremap-breakcore.html) doesn't include "boy band" at all.

Still plenty of surprises... I like enka a lot but I wouldn't have anticipated any of the things it's associated with... "deep filthstep?" http://everynoise.com/engenremap-enka.html

edit: Oh, I'm probably looking at it backwards and those are the least similar genres. "Classic Chinese pop" and various folk musics are less surprising.

Yeah, the top one is the neighborhood, the inverted-color one is the opposite neighborhood (http://www.furia.com/page.cgi?type=log&id=393)...

Is fun just browsing through the genre names.

The scatter plot looks like Denmark.

How has this not been hit with a million DMCA claims and erased from the internet yet?

Looks like it uses an embedded Spotify player, with playlists for each genre. Also, the main page only plays snippets.

Reply on news.ycombinator.com