“Everything About This is Cursed”

Throughout the piece I reused key sounds and processed them in a feedback loop, through the various applications I found online. The last piece ‘ERROR”, features sounds from all the previous pieces processed multiple times over, distorting and warping them far away from their original form. Artificial intelligence for music is still at the infancy stage and is no where near the capabilities of a human yet, but that is likely to change in the coming years.

Almost Human

I used Holly Herndon’s artificially intelligent voice model app Holly+ on this track, which replicates the sound of her voice in a similar way to how DeepFake technology works. I processed the violins and some other parts of the track through the voice app and layered the output back in with the original instruments. The result is definitely artificial and something a human singer would have great difficulty performing but retains human qualities, as it is Holly’s real voice being used. This acoustic/electronic organic/artificial dichotomy is something I have been exploring regularly in my work. The violins are by Francis Moore Colyer.

Almost Human 2

The vocals in this example I created by elongating and re-pitching words and syllables from the audio of an artificially intelligent text to speech bot. The lyrics were generated by the artificially intelligent web application TheseLyricsDoNotExist, which is an application that was trained on thousands of songs in different moods. This application allows keywords or song topics to be inputted and generates new song lyrics based on that. I imagine this song to be the first of many that tell a story of this bot’s birth and development through a series of audiovisuals. For this song, I printed some lyrics with the song topic as “artificial life and death”. I cut out the most interesting and provocative lines and then combined those together into a basic song structure. This method resembles a modern day version of the cut up method popularized by William Burroughs and later utilized by others, including David Bowie and Brian Eno.

I Dream of Machines

This audiovisual piece is a sound collage inspired by Ai art for a masters project. All my classmates reacted musically to some of these images with no restrictions to key, tempo, harmony or melody. I just told them to just go crazy, and they did! Vocals by Jelena Perisic, Eileen McLoughlin on sax and synth, Daniel Ryan provided guitar, accordion and violin, piano is by Karina Moore and Cathy Purcell on percussion. I then rearranged, glitched everything and added some electronic elements. This video shows the images I made using a few Ai apps through all the style-transfer iterations to where it is now at the moment. I used my drawings in these images along with other images including H.R Giger’s paintings. You can spot an alien in a couple of the last pictures. I also put some musical parts through google’s tonetransfer Ai app as another feedback loop, which produced a really wonky, glitchy and melancholic horn sound.

Time Crystals

I’ve been experimenting with @holly_herndon ’s Ai voice model, Holly+ again! This time all the sounds in this track are from the app which I edited, chopped, distorted and bitcrushed in various ways to get those percussive sounds. Dropped an octave and ultra-compressed  the vocals for the bass sounds too.