Experimenting with Music Types in Suno

How many different ways can I not-really apologize for those stupid plums?

Everyone knows about AI-generated images, but AI-generated music is still pretty new. Here’s a quick experiment in the creative range of suno.ai: 10 songs generated with the same lyrics, but different genres.

1. This is Just to Say, Dancepop Version

2. This is Just to Say, Folk Version

3. This is Just to Say, Opera Version

4. This is Just to Say, Indie Rock Version 1

5. This is Just to Say, Indie Rock Version 2

6. This is Just to Say, Lounge Music Version 1

7. This is Just to Say, Lounge Music Version 2

8. This is Just to Say, Experimental Jazz Version

9. This is Just to Say, Sea Shanty Version (If this is any indication, Suno thinks that Imagine Dragons sings sea shanties.)

10. This is Just to Say, Metal Version

Notes:

• All songs were generated from exactly the same lyrics: a spiffed-up version of William Carlos Williams’ poem This is just to say.

• The only thing I changed was the “Music Type” prompt. I didn’t try to be clever or specific: I just named genres: “dancepop,” “folk,” “opera,” etc.

• Suno misunderstood several of these genres. I included these songs anyway.

• A good fraction—maybe a third—of the generated songs were crap. I excluded crap songs.

• In some cases, both of the songs generated within a genre were very similar. In that case, I included only the one that I liked more.

PS: “Folk” is probably my favorite.