Suno Releases Personas – Here’s How to Create an AI Artist

Call, the one AI music production platformjust released Personas. The new feature, which is only available on the Pro and Premier plans, allows users to save the “essence” of a song, such as style and voiceand reuse it again on different song creations.

In fact, it gives creators the chance to build a catalog of tracks that sound like they were made by the same group of musicians, not just a random AI algorithm.

You can even offer yourself I’m calling “persona” a name and image that appears alongside a track, as if the person were the performer of that track.

How Suno Persona works

Flow

(Image credit: Flux/Nigel Powell AI Future)

To generate a Persona, go to a track you’ve previously made and click Create option then select Make Persona. At this point, you can name it, attach a thumbnail image for easy identification, and add a description.

You can also make your Persona private or public as you wish. Once saved, the settings become available for any new song you want to generate.

I tested the new option for a few hours and found some good and some average aspects of the process. The good part was how easy it was to use the new feature once created.

The Personas menu appears in the sidebar, along with a drop-down menu so you can select one of your pre-saved Personas while setting up a new song. Just add or generate lyrics and you’re good to go.

What’s the point in Personas?

Adding characters addresses a big problem with AI music producers like Suno and Udioin that it is very difficult to make two tracks sound like they were produced by the same artist. There are some lengthy workarounds you can use, but in the end it’s still a lottery, due to the random nature of the AI music generation.

But the overall concept is great and something that will dramatically improve the entire AI music creation scene once the technology matures. It will take some time to remove the wrinkles.

Even if you select the same genre details, use a seed, and provide detailed instructions in your prompt, chances are you’ll end up with something quite different from the previous track. The ability to create and store a Persona changes all that. Now you can simply attach one to the track creation process and the platform should do the rest.

Unfortunately, I found the results to be a bit off. Results were decent for clear genres like reggae, but less so for more generic styles. I guess it’s hard to stamp a strong identification on every possible variation of rock or pop, for example.

The vocal consistency of the tracks was also a little weak. Vocal fidelity is roughly consistent if you squint hard enough. But there’s no way these days to create, say, your own Bruno Mars persona and confidently produce a string of bangers using that style.

But the overall concept is great and something that will dramatically improve the whole AI music creation scene once the technology matures. It will take some time to remove the wrinkles. Fortunately, the new Personas option is clearly marked Beta, so it’s probably fair to cut them some slack and applaud their intent.