Want to create your own AI music artist? Here is.


Artificial intelligence is one of the biggest things (if not THE biggest thing) in tech right now. And as programs get smarter and computing power increases, we’re going to find it harder and harder to tell the real from the manufacturer. The whole fake business scares me.

AI researchers seem to be fascinated with how this technology can be used with music. That brings us to Sonya, a fully AI-generated interpreter created by a Ukrainian company called Reface. She (she?) Isn’t the first AI artist, but she’s a proof of concept. With Pacemaker, anyone can create their own Sonya.

I know I know. Sounds a little too Frankenstein /Strange sciences-y, but there is no way to stop it. Britney Spears, Miley Cyrus, and Snoop Dogg have used Reface before.

Here is what we are talking about. The artist and the song are completely the products of AI. An EP titled Love and black glove will be released on Friday.

I quote Pacemaker:

“We used a bunch of different online neural networks to generate speech, sound, and visuals.

Enter a few words to choose a genre and specify your mood, and you will receive your unique name and original lyrics. For example, we used words like love, jealousy, and face. We then put those lyrics in the mouth of an AI voice generator online – and now she has a voice!

The musical part was the most exciting. We used the Boomy Neural Network to generate both the beats and the album art. First of all, you need to choose a genre. Then the network allows you to modify the arrangement and production, change instruments and even add vocals. Finally, you can create an AI-generated song and album art for your release. Just enter a word and get your cover! Want more pictures? No problem. We also used text2art based on VQGAN + CLIP, where you enter a description, choose a quality level and generate an image. Runway ML’s “storytelling machine that automatically generates synthetic images when you write new words and phrases” is also pretty awesome. And Artbreeder is a neural network that allows you to generate a portrait or a character by changing characteristics like age, gender, RGB color values, hair, etc.

Great. Just awesome.

You can read more here.