Can machine-generated music really be catchy? Hello, music industry.
Step One: Let’s Write the Lyrics
We can use ChatGPT, but it’s worth checking out some alternatives, like Claude (still my favorite) or Perplexity (both available in their paid versions).
We can specify the theme, style, verse structure, rhythm pattern, flow, language, etc., if we want—or we can leave it all up to the AI. Chances are there will be some stumbles or imperfections right off the bat, which we can address with two techniques:
Request Corrections – This keeps the previous version intact, which can be helpful if it’s good, but if it’s not, or if there are too many earlier versions, it might just confuse the AI’s thoughts. Sometimes, it’s better to:
Rewrite the Original Prompt – Every platform allows you to modify earlier commands, and by rephrasing or specifying more clearly, we can sometimes make progress more effectively.
Check out some additional tips here to get the most out of language models.
Budapest Dawn - Hit Creation with AI
Step Two: The Music
The unchallenged king of generative music creation is the Suno service. The V4 version is now available, while the video above was created with V3. It’s free to try, but beware—it can be addictive!
While it can write lyrics, its strength lies in music. Much like writing poetry, we can set parameters—and we shouldn’t hold back! It can create music about anything in any style, catering to everything from Hungarian folk songs to any modern extreme requests.
By default, we just need to provide a short description of the song, but in “Custom” mode, we can specify the title, the lyrics we’ve created (which it will sing unless we request an instrumental track), and the style. It typically generates two versions, which can’t be edited but can be continued.
Desert Heat (AI generated with Suno V4)
Step Three: Illustration
If we’re okay with just a cover, we’re all set, as the app also generates an image for it.
Want a video too? Coming up…
