Rhythm and algorithm…

Endel is a Berlin-based company that makes music using algorithms, short pieces lasting around two and a half minutes with ambient background noise bearing titles such as Sleep: Clear Night, Sleep: Rainy Night, Sleep: Cloudy Afternoon or Sleep: Foggy Morning and that can be found on Spotify, Apple Music or Alexa, as well as generated through its app, which uses your location, the time, weather or your heart rate to create custom pieces. I have to say that Endel’s five albums nearly sent me to sleep; more will be released this year, and the company could theoretically produce as many as they wanted.

According to some media reports, Endel is apparently the first algorithm to sign a deal with a major label, Warner Music, prompting some critics to predict the end of music as we know it.

The reality is that Warner hasn’t signed up an algorithm, but with the company that retains the rights to its creations; this is a no-advance agreement with the label to distribute some of its work based on a half-and-half profit share. Endel’s algorithm can generate up to six hundred pieces of music in a single collaboration by combining sounds of all kinds and requires minimal human intervention. It then hires another company to give them titles, while the label simply selects some to distribute through music platforms, sharing space with conventionally created music composed and performed by peoples, generating income based on the number of times they are played.

Media coverage of the distribution deal may have been light on facts, but the event is certainly significant, although record labels have been interested in machine learning for some time, using it, for example, to create playlists on music platforms to promote certain recordings or artists. Nobody should be surprised that the record labels are interested in music by composers and performers who charge little or nothing, who are tireless, loyal and who do try to renegotiate their contract… Did anybody ever really believe that the labels loved their artists?

The potential of algorithmic music is limitless: these ambient compositions are reasonably simple, but a well-trained algorithm could produce to order based on a particular style or using elements from a hit song. In fact, algorithms are already working with musicians, although for the moment they have not enjoyed much success.

We humans are far more predictable than we like to admit and record companies are adept at manipulating the tastes of the market with strategies to control distribution and marketing. We are going to hear a lot more algorithmically generated music in the future, along with a lot more discussion and discord.

(En español, aquí)