Infinity welcomes careful composers: Exploring infinite music systems

Cárthach Ó Nuanáin
The Sound of AI
Published in
6 min readJun 13, 2019

While infinite music sounds like an enticing idea, the skill of the composer or audio programmer remains an ever-lasting necessity. A recent flurry of streaming services delivering endless music promises untapped, yet perfectly adaptable variations. But with the advent of AI-assisted music, composers shouldn’t simply hand what they do so brilliantly over to the machines.

Image taken from http://nautil.us/issue/38/noise/brian-eno-plays-the-universe.

Last week, in collaboration with Dadabots, Melodrive launched a 24-hour Twitch channel of AI-generated music with semantic control. This novel feature allows users to interact and tailor the emotion of endless piano music in the ambient veins of Erik Satie or Philip Glass.

Dadabots, of course, have made numerous newswire headlines these recent months. Their series of albums demonstrates the power of neural networks in recombining curious but stylistically accurate mishmashes of everything from death metal to country — culminating in the entertaining live YouTube channel Relentless Doppelganger.

With server and internet speeds soaring as their associated prices plummet, this recent phenomenon of live-streaming potentially infinite, ever-changing iterations of machine composition represents an exciting, new — and perhaps more appropriate — medium for disseminating and consuming their music. But other projects and artists have experimented with infinite music outside of purely web-based streaming channels. In this post we take a brief look at various different infinite music systems.

Generative music

Digital renaissance man and ambient god Brian Eno reckons he coined the term “generative music” when he teamed up with SSEYO to create the Koan software that outputs music that is “ever-different and changing, created by a system”. Unusually, for a recording artist associated with conventional media such as vinyl, cassette or CD, the result of their endeavours was actually on floppy disk. The intention then was for their consumer to run the software on their personal computers, essentially setting up a “client-side” only stream of algorithmic and ambient music pre Web 2.0.

Eno would later expand these ideas through a series of iOS apps that create generative music in the palms of users’ hands with apps such as Bloom (2008) and Trope, as well as large-scale installation artworks such as Lightforms and Soundforms (Barcelona, 2017).

Software Packaging for Generative Music 1. Image from Intermorphic Systems.

Reactive music

Other artists and developers have also produced apps or artworks that offer potentially infinite musical possibilities. Around the same time as Eno’s Bloom app was released, Michael Breidenbruecker, of last.fm fame, started RjDj in a bid to offer the world a new genre of music termed ‘reactive music’. It’s most distinguishing feature is that users are included in the creative process — not so with generative music. In practice this involved using the built-in sensors from the portable devices’ cameras, mics, accelerometers etc. that pick up information from the environment to interact with parameters of the system. Using the environment to provide input data to interactive musical applications is obviously highly useful in providing variation and novelty. This ensures that the music does not sound too repetitive or pre-baked.

Under the hood, RjDj implements a version of Pure Data to enable developers to design customised interactive music patches for mobile devices. Sadly no longer active, during its heyday RjDj attracted many high profile artists releasing ‘music’ through the device including Hans Zimmer, Carl Craig and Air.

RjDj iPhone interface. Image from evolver.fm

Musique non-stop

One would suppose that a genre of music as repetitive and self-similar as techno would easily lend itself to infinite generation by a computer. In fact, this is exactly what Petr Serkin’s Eternal Flow portable music generator aims to achieve. For $99 you can be the lucky owner of a black box with headphones that spits out myriad iterations of algorithmic dub techno not too far removed from the often copied, yet never equalled sounds of Basic Channel or Deepchord.

Eternal Flow music player. Image from CDM — Create Digital Music.

In practice it’s highly challenging to create convincing computer-generated dance music. The subtle manipulation of rhythmic expectation, arrangement layering, as well as the myriad of mixing decisions means it’s very hard to abstract and encapsulate into a vocabulary and language that is amenable for computer algorithms. Eternal Flow is quite faithful to the distinctly deep and immersive sounds of dub techno, and reportedly performs all the synthesis on the device, which is impressive.

Have a listen here:

Frankie says relax!

We hope you’re relaxing to the soothing sounds of the Melodrive Twitch stream while reading this. If you’re actively seeking more meditative soundwaves, apps like Endel and Mubert make it their business to create customised soundscapes to help you focus on work better or fall asleep more easily.

Endel (yes that Endel) is apparently built by a multidisciplinary team of artists, technologists and scientists and aims to one day “fill hospitality and retail spaces” with considerable backing by Warner Music.

Mubert, on the other hand, claims to be the world’s first generative streaming service offering premium channels at $0.99 per month. Sounds are sampled as opposed to synthesised, which naturally means less flexibility when it comes to variational changes.

Results with these kind of apps vary greatly — comments range from the positive “mushrooms without the mushrooms” to more negative: “at their bleakest, the compositions can resemble goosed-up elevator music.”

Sound art and interactive installations

The allure of and potential for the infinite has not gone unexploited in sound art practices. Previously we mentioned Brian Eno’s large scale installation work that provides endless immersive soundscapes. Indeed, many other artists have also produced similar site-specific works that are limited only by the duration of their commission or exhibition. It’s worth pointing out that works like the Zadar Sea Organ and Panopticon are completely acoustic — they do not use electrical or digital means to create infinite music.

Zadar Sea Organ

In the coastal city of Zadar in Croatia, an organist has been playing and improvising an endless piece of music since 2005, with no sign of stopping. That organist? The ocean. Designed by Nikola Bašić and winner of the European Prize for Urban Public Space, this beautiful work harnesses the power of waves to create random but harmonically coincident sounds.

Panopticon: The Singing Ringing Tree

This is a similar work to the Zadar Sea Organ, but relies on the winds of the blustery countryside in Lancashire,England. This imposing structure chooses not to compose harmonically-pleasing music; intentionally discordant tuned pipes produce an ethereal ghostly soundscape that haunts the sparse surroundings.

Anita’s Dropship

Introduced in 2014 at Burning Man, Anita’s dropship is a “generative dubstep machine that produces drops on demand”. Rather than revellers chasing around the festival in search of “the drop”, this vehicular work will roam the entire festival, until it arrives to greet you spitting out endless dubs — climaxing in a shuddering drop fired off at the push of a big, inviting red button. Full source code including Pure Data patches are available on their Github.

Datamatics

Sonification — the use of audio to convey information or data — is an increasingly common inspiration for sound works, and obviously one that increasingly resonates with us as we continue to become a more connected society. Artists such as Ryoji Ikeda produce large-scale immersive individual installations that draw on endless streams of disparate data sources (maps of the universe, DNA sequences mathematical formulae) to create sound and vision.

To infinity and beyond

With the release of a range of AI-based streaming tools and music services, the potential for musicians and enthusiasts has not yet been fully realised. Hopefully this article has opened you up to the sea of possibilities offered by infinite music that may not rely on internet streaming. Whichever route composers and audio programmers adopt, it’s important that they approach it with care, so that only choose those methods that enhance their craft.

--

--