(Issue 2717, May 11-17 2026)
I found some recordings of an instrumental piece of music called 'Monte Carlo' that I wrote several years ago for a conference. I subsequently added lyrics and melody. I've re-recorded the drums and used the original vocal as a guide vocal for Suno and Audimee. I used this demo for an AI-assisted production called 'The Losing Game'.
2025
A question I now get asked is how much do I use AI? The answer is a hell of a lot. In the past it was all me (Living in a World of Make Believe & Babylon) but now I have a collaborator called Artificial Intelligence. It's actually sound generators that I'm using that have zero intelligence (indeed like all AI software). For years now many music composers have been using loops (such as Apple loops) but sound generators are the next generation.
For the first two albums I wrote the lyrics and recorded myself playing all the instruments to produce the skeleton of a song recorded in my DAW (Logic) (Step 1). I'd then spend months developing and shaping the song and changing and re-recording regions (Step 2). As part of the process I'd put down a guide vocal and then record a singer doing the job properly (Step 3). Finally I'd mix the track and then master it in Logic using some third party plug-ins (Step 4). Now I start as before by writing lyrics and producing a demo song in Logic (Step 1). The difference now is I then put that demo track into a series of sound generators and a vocal synthesizer which I can program to follow the input as tightly or as loosely as I want. I then experiment with multiple ‘takes’ (tight and loose versions) until I have several that I can use. I then extract the stems (vocals, drum, bass, guitar, synth, FX) and put these into my demo project in Logic where I can re-arrange, add more direct recording if I wish, and then mix and master within Logic to produce the final song. So going back to the question of how much ‘AI’ do I use the answer is that the songs I am currently producing would not exist without ‘AI’ but then they wouldn't exist without me either, so it's a real collaboration.
As examples if you were to listen to the 2025 version of ‘He Came Alive’ (with sound generators) and the original 2014 version (me alone with Mark Palmer singing) you can see what a difference it makes. The vocal is uncanny in that it is very similar to Mark but not so at the same time. Similarly my guitar lead on "Living in a World of Make Believe' is clearly derived from the lead break played by me on my Satriani Signature in 2012 but it's a different version in 2025. On this occasion I personally think I ‘beat’ the sound generator on the lead break but usually I don't 😀.
Making the videos is very much AI. I'm trying to make a 'visual partner' to the songs. To do this I instruct image and video generators to make 5 to 10 second clips which I then download into iMovie or Final Cut Pro for editing. The images and videos produced by AI have multiple flaws in them which I don't try to correct as I want them to be seen to be produced by AI and the flaws are part of the appeal. In Rocket Man the astronaut (Mallory Vice) suddenly has three legs at the moment he leaves the ship after blast off and himself turns into a space rocket. In that video there are multiple references to sci-fi TV shows and films that I have always liked. There were an additional two related to Star Wars but Lucas Film told me to remove them……but were very nice about it and wished me well with my artistic endeavours 😊.