The Case for Suno!
One of the most significant stories in popular music this year was the rise and fall of The Velvet Sundown.
To recap, a band appeared from nowhere, racking up enormous amounts of plays and followers on Spotify with no back story, concerts or media presence.
It was of course all an AI artifice.
But it has got me thinking and challenging my own practices as a music teacher, and specifically the teacher in charge of the Y7&8 bands where one of our objectives is to create and play an original song.
Working with a school band of 12 year olds is a lot of fun but comes with many challenges. Putting together a drummer, singers, guitarists,bassist and keyboard player is something I’ve been doing for many years now and it’s never without its wobbles. Mastering a few school pleasing covers is an incredible achievement on its own but composing and performing an original song is a big step forward.
So when a challenge comes up, a Battle of the Bands featuring an original song, where do you start?
For the last 2 years our school bands have been using Suno to generate a song as the start point for our entry. They listen to both Suno versions that came out of the theme and genre prompts they entered. They then chose the best one and started jamming to that.
Both years, the Suno lyrics were largely replaced very quickly with new lyrics, using Gemini as a start point but then substantially reworking the AI version into something truly original.
Over a period of a few weeks, the songs that the bands were playing had moved a long way from their Suno origins. The melody was constantly evolving, riffs and solos were being tried out and added. The creative process was really taking hold and the song became theirs.
Is this use of AI cheating? I don’t believe so and here’s why:
Up until recently, my attempts at getting a band of 12 year olds to write a song went something like this:
Find a tried and tested chord sequence based on chords the band members know how to play.
Come up with a melody based on that pattern- invariably this will be a safe melody firmly grounded in the notes of the chord.
Apply lyrics. Maybe brainstorm some ideas. Try to think of some rhyming couplets.
Work out a feel for the rhythm section.
Make a recording, no matter how wonky so we have something to listen to and build on next time.
In most cases, each of these steps took a lot of teacher input to drive the project forward. At the end we’d typically have another 4 chord song in G (because the guitarist hasn’t mastered F yet), The melody follows the root note and the lyrics are no better than “The cat sat on the mat”.
I’d suggest that Suno does all of that but better and much much faster. It works by taking proven, successful musical elements and putting them together in a slightly different way to create a new work. That’s certainly not a songwriting technique that's unique to AI. That’s a tale as old as time. It’s also identical to the 5 steps above.
The idea that technology has reached the point where we can listen to 2 drafts of a song within 30 seconds of agreeing on a theme and style is mindblowing.
We no longer have to listen to a terrible phone recording of a first attempt at an original song one lesson on, with the kids cringing at their own singing and playing. A Suno first draft is much more inspirational.
The musicians can hear their parts. They can identify the elements which need to be changed and improved upon.
They can hear their own potential in the AI version, and they can transcend it.
That’s where the human creativity element must take over.
Without that our school bands risk becoming a Velvet Sundown tribute act.