Zubin Kanga - Machine Dreams

The Quietus

“We have humanised computers and dehumanised people,” writes Andrea Grimes in her essay The Unbearable White Maleness Of AI. With the non-stop media barrage of hyperbole directed towards so-called artificial intelligence, we have seemingly surrendered in advance to the capitalist agenda. Machine learning tools are here to take jobs, proliferate facsimile art, and obsolete humankind. The medium has finally eaten the message. But what if this technology was like any other kind of technology? Dangerous in the hands of tech bro bastards yet containing immense potential to be unlocked by more empathetic and creative people. London-based pianist and composer Zubin Kanga is just one of countless artists who have spent years subverting, modifying, and tinkering with technology to create compelling artworks and – with a dash of dreamy transhumanism – explore previously inaccessible artistic avenues.

Machine Dreams is Kanga’s most comprehensive work to date. The album is the result of the four-year research project Cyborg Soloists, funded by the UK Research and Innovation Future Leaders Fellowship at London’s Royal Holloway University and commissioned from a number of composers working with bleeding-edge digital instruments. In a sense, it’s a showcase of all the ways in which contemporary technology can exist in a symbiotic relationship with art, from physical interfaces like MiMU sensor gloves to, yes, AI-generated sound sequences. But beyond the oft fascinating technical aspects and compositional ingenuity of the ten pieces featured on the record, it’s their inherent humanity that renders them successful, while sparks of whimsy and authentic emotion give them an endearing sheen.

On the opening ‘Car-Pig’, composed by Alex Paxton, we hear a deranged symphony of cartoon effects packed into sampler keyboards – from alien oinks to wheezy concertina yowls – excite themselves into a hyperspeed, pitch-shifted jazz soundtrack to a Tom and Jerry chase. Even without seeing Kanga play the piece, an irresistible sense of humour and a frenzied sort of glee can be felt in the music. In contrast, Tansy Davies’s ‘Star-Way’ sinks into a calm and earthy vibe, filled with bubbly arpeggios and glistening droplets of sound, which Kanga draws from analogue synths. Meanwhile, Alex Groves’s ‘Single Form (Swell)’ rises and falls along waves of trumpet-like textures, controlled by MIDI keyboards.

Listening to the music and reading about the instruments employed, one can easily imagine the sheer physicality required to perform the pieces. As a pianist, Kanga naturally relies on his whole body while playing, but his works using MiMU motion tracking gloves and similar digital and mechanical paraphernalia elevate the performative, corporeal dimension into a crucial aspect of his art. On Jasmin Kent Rodgman’s ‘One Hundred Random Demons’, found objects, tape players and modular synthesisers become possessed by metaphysical ghosts, then get manipulated alongside light and gesture sensors to create a faux-ambient piece awash in flickers of static and spectral noise.

CHAINES’s ‘Escape TERF Island’ takes this atmospheric malaise and turns it inside out, letting loose a Cronenbergian assemblage of squishy, icky bodily noises, industrial beats, and disembodied choirs. Meanwhile, Kanga’s own ‘Metamemory’ juxtaposes a neural network model – created from his own past recordings of piano music by Alban Berg, Olivier Messiaen, Tōru Takemitsu, and others – with analogue synths and his ephemeral self for gorgeous, pulsing effect. As the final, scratchy notes of the cut fade out, another significance of Machine Dreams emerges. While the music could not have been realised without these specific technologies, it’s a spark of the intangible, of human creativity, that brings it to life. As Holly Herndon recently remarked on Twitter, truly innovative art based on AI will not be found in repetition and echoes of past works but in “approaches we don’t have words for yet”. The music of Zubin Kanga has been heading there for a while.

Share this article: