The Shazam Effect | Hacker News Reader

The Shazam Effect

theatlantic.com

109 pointsnirkalimi34 comments

> “The idea that DJs are just picking songs because they like them is so antiquated,” says Radha Subramanyam, the executive vice president of insights, research, and analytics at iHeartMedia

I imagine that hardly anyone here listens to traditional radio anymore, which makes a lot of sense to me as I'm very particular and selective about my music listening habits too and rarely trust any recommendations.

However recently I started listening to NTS Radio rather frequently. In contrast to the quote above NTS shows are very much made of songs the DJ likes - that is all the qualification they need. No agenda, no hype, no ratings. Every show owner is free to come up with his very own selection, from all genres. Of course it means you have to find the right show for your taste - however if you do it's rather satisfying - I've been able to dig up a ton of rarely known soul gems on "The Do You Breakfast Show": http://www.mixcloud.com/tag/the-do-you-breakfast-show/

The whole idea and philosophy behind NTS Radio is very appealing, worth a read and a fresh contrast to the data-driven selection process described in the article: http://www.dummymag.com/features/the-rise-and-rise-of-nts-ra...

I listen to the radio - classical music stations primarily.

I used to be selective about my music until I felt like I was diving into a computationally created tunnel vision of my own interests. It culturally biased me without me being aware of the bias. I did not like this. I mainly did not like this because I had unhappy things happen in my life, and then all the recommendation engines did was remind me of things I associated with unhappiness.

More generally, I am repulsed by the idea that a formula exists, that defines my taste. I am repulsed by the idea that I am predictably definable. I will happily take recommendations from some place else if I know it screws up an algorithm.

> NTS shows are very much made of songs the DJ likes - that is all the qualification they need

The idea of DJ-as-selector/curator (rather than producer or 'personality') has had a resurgence in the electronic music scene over the last few years. The tastemaker DJs on stations such as Rinse FM and NTS are increasingly reaching wider out or further back to find old gems and weirder music to build into increasingly eclectic sets, and I think this has been driven partly by the mainstreaming and homogenisation of electronic music culture into "EDM" in the US, but also as an antithesis to the machine-made reductiveness of Spotify genre mixes.

As a regular club-goer here in Europe, it's no longer enough for a club DJ to merely play one genre all night long -- a crowd expects to be surprised and entertained.

If you're looking to widen your scope, the electronic music community has some great resources out there, and you can dive into any of these and find something utterly new and exciting without much trouble:

1. http://www.residentadvisor.net 2. http://www.xlr8r.com 3. http://rinse.fm 4. http://ntslive.co.uk 5. http://musicforprogramming.net

Resurgence? This was the DJ's role in the electronic music scene going back to the 90s, back when everyone spun vinyl. Just saying this isn't new at all there.

Agreed. Sure, as I have been around when mp3s became a thing and the TranceAddict or MercuryServer forums where the go-to place for electronic music enthusiasts, I'm of course familiar with your sources. In fact I produce and to a lesser extent DJ myself.

Of course since then my taste became much wider and genres are really irrelevant to me today. Which renders the usual recommendations engines hardly useful to me - it seems everyone's coming up with the same suggestions (or they all use EchoNest). In any case it's not a replacement for a real human curator that knows his stuff. Also labels play a pretty big role in this. Despite the evil reputation of the major ones, their smaller siblings are very valuable when it comes to curation and selection. Being featured on kranky, warp, Secretly Canadian, Editions Mego,... well, you name it, is a very strong indicator for quality. Particularly in electronic music this is hardly news.

Speaking of digging up hidden gems and rare classics: Do you know the Numero Group? An utterly fantastic re-issue label that made it their mission to find the most well hidden and unreleased material to give it a proper modern release. The background story is a fascinating read too: http://www.theguardian.com/music/2013/jun/13/numero-group-cr...

If you want to learn more about them, start here: http://m.mixcloud.com/NTSRadio/numero-six-hour-takeover/

I think people misunderstand how & why pop music tends to be simple and banal, relative to more grown up genres. This is kind of tangental to the article, but somehow always part of the discussion about hits. People usually blame labels and such, risk averse and uninspired. I think this isn't the real reason.

Take The Beatles and all the British rock from the mid to late 60s, Stones, Zepelin, etc. They basically took American "delta blues," a rich, but fringe musical tradition and they played their simpler version of it. They weren't (especially at the start) the same level of musicians as the original blues guys. They played it as well as they could, a simpler version. It came without the racial baggage & cultural biases. The result is a more digestible version.

There's a version of this story for Elvis, and lots of other pop icons.

The musical and cultural palates of suburban teenagers couldn't digest bluegrass, boogie boogie, delta rules or other mature folk sounds. They could digest the beetles, a simplified version blended with the pop music sounds people were used to.

The Beatles' music matured and fanned out as did the whole scene that came with them. People could take the harder sounds that they adopted. Rock became its own mature genre. I don't think Kashmir would have sounded as good in 1963 as it did in 1973, before Clapton, Jeff Beck and the other artists from that movement gradually built that guitar style on top of a simplified blues base.

That's what pop is. It's an accessible version of music. Boy bands sing harmonies that you could find in 19th century music that kids would hate. They simplify it, sprinkle it with sounds that kids are used to and package it in my-older-brothers-handsome-friend form.

We see this with a lot of indie music. An eclectic mix of folk music. Django guitar and ella fitzgerald vocals. Its repackaging music for new listeners. Teenagers are new to everything so this type of music works well for them.

It's not a bad thing and its not an insidious plot.

I think you're generally right about the fact that pop music popularizes more grown-up genres.

But I think you could have picked something different than delta blues as an example. Delta blues was extremely simple when it started out, before the end of the thirties there wasn't even a proper turnaround [1]. If you compare originals from before WW2 to covers by British bands you'll find out that in most cases the British version is more complex.

On the other hand take Frank Zappa's Freak Out and compare it to Sgt. Pepper's…

[1] https://en.wikipedia.org/wiki/Turnaround_(music)

I was going to go with Prince & Funk. I think there's a similar thing going on. But simplifying a mature & complex style, is only a subset of what pop music does. In Princes case, he made the music more complex but also more lyrical and otherwise more digestible.

Related: Creating Shazam in Java http://www.redcode.nl/blog/2010/06/creating-shazam-in-java/

HN discussion: https://news.ycombinator.com/item?id=5723863

More Shazam-related: https://hn.algolia.com/#!/story/forever/0/shazam

My company[1] is a similar service for the artisan/premium beverage industry (focusing on coffee, beer, and spirits). We use machine learning and data science to understand what individuals and populations taste in a product, what attributes they're selecting for, and what they like and dislike in the product.

We then use that data to monitor their products in real time for quality control problems (flaws, taints, contamination, and batch variations), and create production and flavor profile optimizations.

I think Shazam (and my company) are offering a valuable service to artists and creators, who never had access to these types of optimizing and feedback tools before. At some level, music and cooking is an art - but its also a business, and the art is based on appreciation and consumption; and thus feedback, assessment, and optimization are critical.

I believe in the future, both music and high end drinks will be created on a near individual level - that they will be targeted and modified and optimized for very small groups and demographics. I believe that is a good thing for everyone; better products, higher quality, made for me, and that its tools like [1] and Shazam that are making this possible.

[1] www.Gastrograph.com

I don't know... what you're talking about is something like how Netflix used their data to motivate shows like House of Cards and Orange is the New Black, which they "knew" in some sense would have mass appeal. The masses are easy to please because you can just average over their individual tastes and capture huge swaths of the general population (see also: political parties, popular music, "middle-brow" chain restaurants), but it's much more difficult to get an accurate prediction for a single person. If you zoom in close enough, everyone becomes an outlier.

Also, creating art is not as simple as entering the infinite realm of possible songs (or paintings, or meals or...) and just choosing a set of parameters. Again, a broad set of parameters, sure. I'm going to write a rock song. I'm going to cook a boeuf bourguignon. But how to capture the weird unsystematic idiosyncrasies of any given individual, and do so in a way that doesn't feel soulless and artificial?

No, Netflix uses collaborative filters (not feature learning) for recommending shows, and does not collect feature data on individual "consumer" preference (at least not publicly).

What my company does uses sensory and preference data at the individual level, and projects for demographics and populations. Good food doesn't feel fake. Good drinks don't feel fake. In the long run, few products will be made by averaging tastes for the mass market - that's what coke and pepsi are.

He isn't talking about "recommending shows," he is talking about how Netflix used data science to determine what shows to produce

There's something infuriating about record labels figuring out if something will hit and then jumping on it.

In the record label golden years, record labels could justify their outsized piece of the pie with the idea that (a) the make stars and (b) they take all the risk. These days, digital recording and distributions makes it all cheaper and de-risks record production. This kind of thing de-risks things further.

If you don't think a record company would are add enough to justify their share, sign with someone else or keep doing it all yourself.

So now we wait for somebody to come up with a way to fake ratings on Shazam (eg, by using a botnet) and watch a mediocre artist become a sensation? :)

https://xkcd.com/1192/

As someone who literally listened to the same tune on repeat for a month I don't see the problem. It was a repetitive, low-dynamic range tune by the way. Perfect for drowning out the noise of the nearby construction work!

If it disappoints the music PhD's that every song sounds the same, maybe they could try to precisely quantify the perfect rhythm imperfections.

Shazam is a mystery to me.

I heard of it and thought wow, when I have a song stuck in my head but don't know the name, this App would help me to find it.

But it only worked if the real song is playing right now.

When that is the case, I'm either in a club and can ask the people there what it's called. Or I'm on a video/music streaming site where the name in 99% of the cases is written out.

Okay, sometimes I have a wrongly tagged MP3 or hearing some song in the background of an ad that I like to know the name of, but that happens only <4 times a year.

Are my music-listening-habits so different from the rest of the world?

Where do most people listen to music without the possibility to get the name of the work?

Sometimes it's easier to open up an app and see what's playing then go up to the coffee shop employee and ask them what was just on the radio. Then if they don't know, yell to everyone there if anyone knows what song just played.

haha, yeah true...

Funny thing is, after writing this post, I remembered an mp3 I had long time, which I didn't know the artist to and shazam found it after 3 seconds listening.

www.midomi.com and soundhound also works by humming or singing the song, I don't know why shazaam hasn't integrated this into its features.

I use Shazam most often in bars and restaurants.

Hmm. So if I'm reading this right, Shazam makes their money from user data? I've used Shazam in the past and never realized that the data was being aggregated and sold. Interesting.

I guess the argument is that the service is free and provides value, so the data can be anonymized and sold?

As far as music goes, this seems backwards to me - and ripe for gaming. It feels as if artists are trying to score some sort of Shazam effect, rather than trying to produce good music and let the chips fall where they may. Music should be an art before a science.

There's probably some license boiler plate on the product somewhere that lets them do whatever they want. No argument needed.

> but is it bad for music?

How about we work off the premise that everything the music industry does is bad for music?

In the article they mention how robots drumming the perfect beat is off-putting, or less desirable to a listener. The current state of affairs in the dance music world is exactly opposite of that. Most songs are created using programs such as ableton, cubase, FL studio etc. As such you can create a 'perfect' song that will do well.

Creating songs with machines/software is different than having machines themselves create the songs. Or what are you saying?

> There is no evidence that Led Zeppelin’s “Stairway to Heaven,” one of the most famous rock songs of all time, was ever played on the radio in the years immediately after its release

I can confirm that it did in the early 70's - I recall it because it was one my parents didn't change the station on.

As someone who doesn't listen to "hits", I fail to see the downside.

Hits already did much much worse things to perception of music than shazam-isation may ever.

Is there a paper that describes the algo?

Yes, there is: http://www.ee.columbia.edu/~dpwe/papers/Wang03-shazam.pdf

It's a quite good read.

has anyone implemented an open source equivalent? It would be useful to see some code

edit: found a MATLAB implementation: https://github.com/LNSD/LNSD-ShazamMatlab

Also, try echoprint

Reply on news.ycombinator.com