Online Music Discovery - Here's An Idea...

[Guest Ralph Nolte]

I don't know about you, but Google searches like "sounds similar to ..." or clicking through random playlists can waste hours as one tries to find the good stuff hidden behind all the "curated content"! Forums like WATMM can obviously point to hidden gems etc. but no two people have the same taste in music - so how to discover tracks customized to YOUR tastes?

 

Modern tech hasn't really come up with a good solution to this, so here's my idea for something I'd like to see one day - let's call it the Google Earth of Music (GEM):

A user would start searching on a colour-coded graphic that shows the major divisions of music currently online. The largest blocks in primary colours are the broadest categories (classical, rock, pop, electronic, etc). Sub-regions in more subtle hues gradually take you to sub-genres and individual artists - right down to single tracks which are sorted algorithmically and clustered according to similarity of song characteristics (bpm, melodies, timbres, compositional structure, length, etc). You can zoom in on an area until you reach individual cells which contain preview tracks. Or you upload a favourite sample and get directed to a "results region" where the original song or similar ones are grouped together. Once you've found something you like you can either listen to the full track or purchase it.

 

GEM might collate music from all over the Net, rely on musicians submitting their tracks and/or be based on paid inclusions for set periods. Each eligible track's waveform is analyzed to assign a specific position on the "content grid" where searchers can home in on it as described above. Metadata can be considered separately, so conventional text searches can also turn up specific artists's work. Popularity rankings play no role in the layout - if something sounds like Ae or Afx, it'll end up positioned amongst their stuff: an even playing field for producers and consumers.

 

What do you guys think of such a concept? Alternatively, what currently works best for you to discover new music that you actually like?

 

[percy roger smith 2]

http://everynoise.com/engenremap.html Have you looked at this? The genres are kind of shuffled because the engine purely sorts by density and rhythm of waveforms but it's what I use to find new artists

[Guest Ralph Nolte]

That's a great find - thanks! I hadn't seen this before, but it looks like a good prototype of what I have in mind...

 

It seems "EveryNoise" has been online a couple of years already and is a proof-of-concept type project based on Spotify data only. The coding sorts according to Spotify's own defined genres (some pretty ridiculous names there, I see) and then picks One artist per sub-genre > One track by that artist > a 30 sec sample of that track to preview. So this is really the tip of the iceberg of something that might work on a much larger scale... Obviously, that would require some serious funding and back-end technology to work as a music search engine for the masses. I do think that this is the way to go in the future - how music is currently presented to users is just terribly flawed, seeing as it's based on ever-more convoluted genre descriptions and marketing spin that puts listeners in silos where they aren't aware of alternatives to the popular stuff they mostly listen to.

[Guest Ralph Nolte]

BTW - Riverdance is  a nice track! 

[Soloman Tump 874]

You could also use this to find gaps in between other genres that do not exist but should.

 

For example there is a huge gap between complextro and industrial metal.

 

Make it happen.

[iococoi 2444]

http://everynoise.com/engenremap.html Have you looked at this? The genres are kind of shuffled because the engine purely sorts by density and rhythm of waveforms but it's what I use to find new artists

 

deep discofox..dear lord