As warned, and also, Why Aren't You Getting Some of This?

Explore how Machine Learning and AI can expand musical creativity while keeping the human in the creative workflow. This forum is dedicated to respectful dialogue where diverse perspectives are welcomed.
RELATED
PRODUCTS

Post

BONES wrote: Wed Jul 01, 2026 2:41 pm We've just finished an album where we started by generating around 400 songs in Tunee, which is a music AI. As we were working on them we selected songs that met our criteria and I spent a week or two on each, deconstructing the song and rebuilding it in Studio Pro. After about three months we had enough good material for an album. Shortly after, Tunee lost its contract to use Suno's models and the models it started using were pretty useless, so we were lucky to get things done when we did. (Tunee is free, Suno costs money.)

The album has 12 tracks, two of which are short interstitials we made ourselves. The 10 songs we wrote with Tunee have varying degrees of AI sound still in them. 3 or 4 of them have absolutely nothing left from the AI generated original and 3 others have backing vocals from the AI version. Some of them have Battery augmenting the AI drums (I used stem separation in Studio pro to split the tracks), either the drum track as the AI made it or a loop made from the drums. We also left a few spoken word voice snippets in a few of them and a couple have bits and pieces of synth parts from the AI left in to fill out the sound a bit more. I've also added extra parts to some of them and left parts out of others. If I had to put a number on it, I'd say that less than 10% of the sound you hear on the album was made by AI but around 80% of the music was written by the AI, under our direction. We fed it most of the lyrics, although it sometimes changed things around a bit.

For me, the process didn't feel significantly different from working with my bandmate's original ideas. It was really like having a 3rd member of our songwriting partnership. We all had our input, in order - Craig (my bandmate) first, then Tunee, then me, then back to Craig and lastly to me for final mixing and mastering.

From go to woah this album has taken us maybe a month longer than the last one took, so we weren't doing it to save time, and it is without doubt the hardest we have ever worked to get an album finished so it definitely didn't make the job any easier. What it did was make the songs we've come up with a whole lot better than we'd have done on our own and it pushed us out of our comfort zone, it made us try new things we'd never have thought to try otherwise. It is also the first time that I'd say I actually enjoyed the process, which is why we worked so hard on it.
noiseresearch wrote: Wed Jul 01, 2026 10:15 amPS: I leave any ethics beside as we now could ask where all the AI is getting the "ideas" from.
From the same places we do. The music I was making before I discovered the EBM genre is very different to what I've been doing since I discovered it. It's had a profound influence on the sort of music I make. And we're all the same, we are a product of our influences. Ai doesn't have the luxury of growing up so it gets its exposure to these influences from its training.
Hi Bones, thank you for your detailed explaination, very intresting and insightful Mate! Hope yourAlbum does well in your genre and community!

Post Reply

Return to “Machine Learning and AI for Music Creation”