Wednesday, 7 May 2025

Proper Trance or Proper slop?

My two favourite genre/eras of electronic music are Italo disco circa 1983/4 and Goa trance circa 1995/96. When Suno v3 dropped in April last year I made some pretty convincing retro-sounding Italo tracks straight away (e.g. this) but, until now, it couldn’t do “proper” trance. Early versions of the model didn’t understand the parameters of the genre properly or generate sounds of a high enough fidelity for my liking. There is a fine line in trance music where tracks can easily become cheesy and most of the things I made with previous models fell firmly on the wrong side of that line. With the newly released v4.5 that’s changed, perhaps partly due to the fact that the model can now make up to 8 minutes of music. A new era indeed.


Listen to "Proper trance with Suno v4.5" playlist on suno.com


Yesterday I uploaded a ~2 minute unfinished track that I composed over 20 years ago (at the time heavily inspired by the likes of Hallucinogen, Prometheus, Astral Projection, Process, Etnica/Pleiadians and X-dream. Before breakfast this morning I had generated 8 complete tracks (using the suno cover feature) that I think, for the most part, hit the nail on the head. They all have the atmosphere and vibes of my original track but are, frankly, much better.




Goa trance music and the way it uses creative sound design and highly technical studio production have played a large part in influencing my career and many of the things I have made over the years, e.g. Beat Morpher (2004), Numinous3D (2018). 


The contentious ethics around how generative AI music tools like suno were created and the negativity around them is quite frustrating. As a music technologist I find them fascinating but I've heard so many people say "that's just not interesting to me", and this surprises me. I think many people focus too much on the text-to-audio functionality and not on the audio-to-audio functionality, which is a lot more interesting and personal for music makers. Perhaps people look at the lowest-common-denominator garbage on the front page of the suno website and are immediately turned off, or they don't play with it enough to work out how to generate music that appeals to their sensibilities. They hear yet another song with a lyric about "neon lights"... etc (suno's newest lyric generator is a lot better!).


For me, I find using suno engaging. It brings a lot of joy to iterate on tracks, to filter good from bad, and has got me engaged with sound in the studio like I haven't been for a very long time. Over the last year I've generated over 6000 suno "songs", and there are around 100 that I've selected/curated, often after many iterations. Some of the best of those are based on my own music that I made a long time ago. Fortunately I have given up on my dreams of being a trance star, so I really don't mind what happens with these half finished bits of music that I upload. You might want to read the suno TOS before doing that with your own composition.


Although I enjoy using the tool, I just use it for entertainment. I feel awkward about the fact I am given the rights with a pro account to do what I like with this music, including selling it - considering the training data issues, this interview, the awfulness of big-tech gobbling up the creative industries etc.  As long as I keep the music I make on suno or label it clearly as AI music my conscience feels mostly clean. Now that you can edit the artwork on your Suno profiles, I don't mind just keeping it there, but I have uploaded a few things to Youtube, soundcloud etc - labelled as AI generated.

I really hate the fact that bad-faith actors are trying to pass off AI generated music, often made with no curation as real music. If you visit the Apple Music page for the legendary trance act Astral Projection, you will find several weird AI nonsense tracks that are not theirs. The fact that anyone can upload anything they make with suno etc to multiple DSPs and that soon they will be totally overwhelmed with AI content is awful. Although I think this technology will be an inevitable part of future music production, right now I would be happy with a blanket ban of AI tracks on Spotify, Apple Music etc. There are many AI detection services, and the models are supposed to be watermarked, so this should be possible. Whether the companies that run these platforms want to do that or not is another matter, but from Apple at least I would expect better.


https://music.apple.com/us/artist/astral-projection/135692863








Saturday, 15 October 2022

iPlug2 is awarded 25K grant from Google!

Very happy to announce that iPlug2 has been awarded a grant of $25k from Google's Advanced Web Apps Fund! 

https://developer.chrome.com/blog/advanced-web-apps-fund-oct-22/

Thursday, 14 January 2021

TheAudioProgrammer Talk on iPlug2

Thursday, 3 September 2020

Music Hackspace talk on iPlug2

Saturday, 27 April 2019

Casio Banks for Arturia's CZV

I was recently involved in the development of Arturia's CZV - a brand-new CZ emulation which is part of V Collection 7. It sets a new standard in emulation and adds a tonne of modulation capabilities and great sounding FX, with a fantastic user interface. The V Collection has an excellent preset management system, and I've taken the time to convert and tag the original Casio patches from the CZ1, CZ101 and CZ230S,  as well as the RC10, RC20 and RC30 banks.

Look out for those classic sounds like SEA GULL...



CZV: OUT NOW - > https://www.arturia.com/products/analog-classics/cz-v/overview

CZV CASIO BANKS -> https://olilarkin.co.uk/downloads/CZV_casio_banks.zip

Tuesday, 4 December 2018