I wrote the lyrics from scratch and composed a crude demo of the song in Renoise where I even sang with my own voice.

I then imported it into Suno and allowed it to improve upon it by making a more professional arrengement and made my voice sound decent. This required about 100 generations until I had all the snippets and backing vocals I needed.

I then downloaded the stems and worked for about a week in a DAW to arrange everything according to my taste. At the end I removed a complete verse and moved some text around to make the song at a good length.

I did the mastering manually and sent it to be published.

Since this would take 2 weeks I used that time to make a music video. Just for fun I wanted to see how far it was possible to get with only using free tools so all the settings and surroundings are images created by ChatGPT. I then imported them into Grok to create the animations. The woman in the video is a digital copy of my wife Susanna and even though her appearence varies some from clip to clip it is scary close to her real looks at times. The man in the video is a complete imaginary man created by Grok with no resemblence to any real human.

Once I had all the video snippets I needed, I created the video in DaVinci resolve which took me about 4 days.

So all in all, about 4 weeks of full time work to make it. Sure, I had the help of AI but in no way is the song written by AI. It’s been a tool, a helping friend, a collaborator.

This is NOT a prompt song where Suno did 99% of the work.

I saw a way of combining Euphoria by Loreen with Stardust by Jean Michel Jarre. It took me some time but I think the result is rather good