RNCM PRiSM
Robert Laidlow (2022): Silicon
00:00 - Intro
00:27 - I (Mind)
16:52 - II (Body)
29:36 - III (Soul)
Recorded live at the WP given by the BBC Philharmonic & Vimbayi Kaziboni, Bridgewater Hall, Manchester, 29/10/2022.
Part of PRiSM Future Music #4:
https://www.rncm.ac.uk/research/research-centres-rncm/prism/prism-news-and-events/future-music-4-25-29-october-2022/
Producers: Matthew Bennett & Robert Laidlow
Recording Engineer: Stephen Rinker
Live Electronics Cues: Bofan Ma
Videography & Post-production: Thirdman Productions
-
Programme Note:
Silicon is music about the orchestra, about technology, and about what an orchestra might be in the age of rapidly advancing artificial intelligence (AI). It is also music that uses technology; AI permeates this piece from start to finish, from top to bottom, as composer, instrument, performer, improviser, and inspiration. Like all technologies, AI reflects the values and interests of the society that creates it and throughout Silicon it primarily acts as a mirror through which I re-examine the orchestra.
This piece is also a tribute to the symphony, although I don’t know whether that makes it a symphony itself. It does have some of the trappings of an 18th-century symphony. However, if Silicon is a symphony, it’s one for those who grew up in the Internet age like me - an age of social media-induced numbness, information overload, a vanishingly uncertain future, and an overbearing, immutable past that can often feel oppressive in its ubiquity. Like the Internet, the orchestra contains near-infinite possibilities, a deep history, and many biases.
The first movement, ‘Mind’, relates to the future and the past. Often, the benchmark for a ‘successful AI’ is how convincingly it can copy something that already exists, such as a Beethoven symphony, a Rembrandt painting, or someone’s face and voice. This movement uses an AI that creates musical scores. Almost always, these come out in the style of an
existing (long dead) composer, but with uncanny twists: sometimes the music will suddenly begin travelling backwards through time, to obsessively repeat over one idea, or to begin referencing itself. I wondered what this says about the music that orchestras play, and the way that AI is designed today. As a living composer, the last thing I want to do is recreate what already exists. This movement, therefore, tries to make something new out of something old. It begins with simple material played by the orchestra, composed in turn by myself and AI, but quickly falls down a rabbit-hole to somewhere both utterly alien and inescapably familiar.
Movement II, ‘Body’, is primarily concerned with fakeness, authenticity, and control. AI is used constantly to create fake, or ‘deepfake’, content online - I wanted to explore the idea of fake music, which is both funny and unsettling. There’s a prominent AI-powered instrument, made specifically for this performance, morphing between the sounds of the human players.
The piece spins between several deepfake dance styles at the speed of a social media algorithm with a disappearing attention span, but the orchestra, like a marionette, is always beholden to the AI lurking in the background.
‘Soul’, the last movement, is the most direct example of AI acting as a mirror for the orchestra. Specifically, it acts as a mirror for this orchestra: the BBC Philharmonic. We hear, alongside the physical players, the sounds of an AI that has listened to and learned from decades of BBC Philharmonic radio broadcasts. AI does not distinguish between ‘sound’ and ‘music’ in the way that humans do, and so in amongst the eerie, brash, beautiful, and lyrical AI-generated orchestral sounds are the phantom voices of radio presenters, the solemn act of tuning up, and the applause and cheers that it has learned from you - the audience. To me, these sonic rituals are the fingerprints of the BBC Philharmonic and get to the heart of
what an orchestral performance really is - an act of community. Perhaps one future for this technology is its inclusion in and contribution to this community.
-
AI tools used during the composition process:
MuseNet: https://openai.com/research/musenet
FolkRNN: https://folkrnn.org/
Google Magenta DDSP: https://magenta.tensorflow.org/ddsp-vst
PRiSM SampleRNN: https://github.com/rncm-prism/prism-samplernn
-
Acknowledgement:
This work was composed as part of Laidlow’s doctoral research (“Artificial Intelligence in the Creative Process of Contemporary Classical Music”) at the Centre for Practice & Research in Science & Music (PRiSM) at the Royal Northern College of Music. This research was made possible by a Collaborative Doctoral Award in partnership with the BBC Philharmonic, supported by the North West Consortium Doctoral Training Partnership, the Arts & Humanities Research Council, and the National Productivity Investment Fund.
The project & its documentation were also supported by PRiSM through the Research England fund Expanding Excellence in England (E3).
