MACHINE LEARNING AND MUSIC AURAMACHINE RESEARCH TRIP
BRITISH COUNCIL | FEBRUARY 2020
As part of British Council’s UK-RUSSIA year of music programme I travelled to Russia to undertake creative research for AURA MACHINE, my (then) speculative project idea, a proposed artwork in the early stages of development to create a sound sculptural machine that uses machine learning algorithms to generate new audio output based on a data set of material concrete sounds. I wanted to explore the potential of live inputs with neural networks to realise future materialities. Supported by my UK partner FutureEverything and RU partner Ilia Symphocat, the main aims of my trip were to
- Connect with artists using machine learning in their artistic practice
- Collaborate and perform with my host part Ilia Symphocat
- Sound Sculpture and machine design – explore visual aesthetics through visiting collections of technical and sonic objects
- Field Recordings – recording urban sounds for ML datasets in St Petersburg and Moscow
I arrived in beautiful St Petersburg to meet my host partner Ilia Symphocat, ambient composer, Simphonic Silence Inside label owner and curator of the Sound Museum to learn of his experiences and perspectives of performing with AI. On the first night Ilia took me to a DIY space for local artists for an exhibition opening from Nikita Panin, an artist working with visuals and machine learning. Nikita took inspiration from the inside of disk drives and considered the synergy of forms with Russian orthodox spiritual iconography. He trained a neural network on a dataset of 6000 sacred images and abstract paintings, presenting the outputs as large scale videos, prints and sculptures. The aesthetic of this was vibrant and pretty psychedelic and through speaking with Ilia’s artist friends, I learned that this visual style was quite typical of 90s St Petersburg rave culture. It was great to gain an insight into the local experimental arts scene and meet artists and musicians who were so welcoming and interested in our arts community in Manchester (especially Joy Division!)
I spent time at the Pushkinskaya Art Centre 10, a legendary alternative complex housing many artist studios, museums and galleries including the wonderful Sound Museum and Museum of Nonconformist Art. I met curator Lora Kucher who took me around the current digital photography exhibition she had worked on featuring Manchester and Russian artists and spoke about their work and history.
My host Ilia lived in the Pushkinskaya building having his studio there and working as the curator and events manager for the Sound Museum. Following some time recording objects in the museums collections, Ilia and I spent the day at his studio, talking and sharing our experiences of making music, performing and about our cities and music communities in Manchester and St Petersburg which incidentally are twinned cities. Ilia shared his experience of working with AI in live performance, he had worked with datasets of classical and jazz music which he reworked and improvised with in real time for his live set at Gamma Festival. We discussed how ‘live’ this can really be and the complexity and perspectives of working with this high level of technicality when you are a musician, as well as the importance of collaborating equitably with technologists.
We performed at the Museum of NonConformist Art , a highly improvised and collaborative experience which we both enjoyed immensely. Blending granulated field recordings and freezes with radio transmission fragments and some live objects. We had a good attentive crowd for a noisy Sunday evening in the gallery with Ilia hosting and introducing me and interpreting for the audience while I talked through the electronics and DIY interfaces I was going to play with. We worked with live visual artist Mikhail Mesyac who created a live digital backdrop to the set.
Other activities in St Petersburg included research on sculpture design through visiting the ‘Kinetic Art in Russia’ exhibition at the Grand Exhibition Hall which was hugely inspiring, charting it’s origins in constructivism to artists working with electronics in the present day, field recording around the city and with objects at the Sound Art Museum and working on music in my temporary airbnb studio! I found an amazing radio at the flea market on the outskirts of town, a Russian 303 with no aerial simply emitting white noise. I used this object as part of my performance with Ilia on the Sunday, combined with my field kit exploring and sampling the broken frequencies of each area I travelled to. It was my first time in St Petersburg and I loved the elegant city and the sparky people who were so welcoming and generous.
I was excited to travel back to Moscow, it’s been five years since I was first here with Noise Orchestra, Dave and I had been researching 1920s noise machines and playing and making electromagnetic field recordings around town. I took the bullet train from St Petersburg very early in the morning, taking in the terrain between the two cities and pulling into Moscow, I forgot just how bombastic it is, the sheer scale and might of it. I enjoyed riding the garden ring at night and taking in the dramatically lit architecture and boulevards stretching off into the distance , I also remembered walks through Gorky park and quiet times recording on the streets around Tsvetnoy Bulvar and Patriarchs Pond.
I spent time with artist and curator Helena Nikonole and creative ML musician and technologist Nikita Prudnikov who are frequent collaborators on a range of projects exploring art and machine learning. One of their most recent collaborations Bird Language explores the structure of bird sounds through experimenting with neural networks. I had met Helena previously in Berlin on ‘The work of art in the age of artificial intelligence’ project at CTM and Helena had led the AI Hacklab for Gamma Festival. I was interested to discuss their experience of live sampling in AI, practicalities of datasets and their applications (for example soundscapes v one shot sample libraries) and potential open source methodologies.
I learned a lot practically and conceptually from Nikita and Helena around different neural network architectures from GANs to Auto-encoders and types of dataset. Nikita also shared a number of sounds and pieces testing these architectures some including voice, a call and response with an AI that was totally uncanny and quite eerie, a style transfer method featuring Brodski and footage of a live exhibition (Helena had curated) where he had performed with Katerina a harpsichord player. The sounds were unlike anything I had heard before and I loved the concept of working in latent space to seek timbres. I spoke about my work with radio frequency searching recordings and we discussed the parallels here with exploring space and working with fragments. It was also good to talk with Helena about our experience and the importance of genuine collaboration between artists and creative technologists when working in advanced fields such as this.
Huge thanks to British Council for enabling me to undertake this trip. It was so inspiring to get to spend time with such amazing people, build connections between Manchester and St Petersburg/Moscow and learn about this field. In terms of my practice it has helped immensely with thinking about the next steps in project development for AURAMACHINE. Special thanks to Irini Papadimitriou and the FutureEverything team who supported me as my UK partner, and Tom Sweet and Evgenia Gerasimova from the British Council for their wonderful support.