Is AI-generated music going viral next?

You’ve heard about AI writing articles, essays, and viral LinkedIn posts. But what about music? While the topic might strike a chord (pun intended) for technophiles, AI-generated music, like many of the other areas AI touches, can become litigious real quick.

A fun tool that generates Drake songs about anything in under a minute launched a couple of days ago. drayk.it uses GPT-3 to write the lyrics based on the prompt you give it. With the help of voice synthesis and “some music magic sauce,” it generated a song about launching on Product Hunt.

While ChatGPT and DALL-E were all the rage this past year, let’s not forget about OpenAI’s MuseNet from 2019. The deep neural network powered by GPT-2 generates 4-minute musical compositions with ten different instruments and can combine styles from country to Mozart to the Beatles.

Google’s been a player in the space, too. While we might’ve not heard much about Magenta Studio lately, last week the Google Research team published a paper on MusicLM. The model generates 24 kHz music from rich captions such as “The main soundtrack of an arcade game. It is fast-paced and upbeat, with a catchy electric guitar riff. The music is repetitive and easy to remember, but with unexpected sounds, like cymbal crashes or drum rolls.”

We’re about to see a lot of debate around IP laws. Who owns machine-generated music? Those who write the code and process the data? Those who write the prompts? As Robin Thicke would say, “I hate these blurred lines.”

Let’s discuss.

–