Spotify's AI DJ, 'X,' is a human voice that says, "It's a digitized version of me."

Spotify's AI DJ, 'X,' is a human voice that says, "It's a digitized version of me."
Spotify's AI DJ, 'X,' is a human voice that says, "It's a digitized version of me."

Xavier "X" Jernigan is the model behind Spotify's AI DJ, which sounds more human than robot.

Spotify's head of cultural partnerships, Jernigan, is also the voice behind DJ X, the music streaming application's AI-powered DJ.

The feature employs AI technology to analyze users' music preferences, mood, and listening habits to provide a personalized stream of similar and new songs they may enjoy, along with their favorites.

The tool, launched in the U.S. and Canada in early 2023, not only creates custom playlists but also communicates with users and offers interesting facts and personalized commentary about the artists and genres they listen to between songs.

Jernigan is the AI DJ's voice that can make users feel like they have their own personal DJ at their fingertips.

Jernigan explains to CNBC Make It that it's a digital version of him recommending songs and providing context and storytelling through AI technology.

Becoming Spotify's AI DJ

Jernigan had previously hosted several original podcasts, including Spotify's first morning show, "The Get Up," which attracted over 6 million listeners before its cancellation in April 2022.

Jernigan was approached by Spotify's head of personalization a few months after "The Get Up" ended, regarding a new product their team was developing - an AI-powered DJ.

Financial wellness leads to happiness, wealth, and financial security.

The AI DJ's placeholder voice was not satisfactory, so the team sought out a voice model to make it sound more like X.

Jernigan says that what he loved about the conversation was that it focused on forming a partnership between him and Spotify, and that the resulting product would reflect his personality, not just his voice.

Jernigan's previous experience at major music labels, including Def Jam Recordings and Republic Records, enabled him to bring not only his passion for music but also valuable industry knowledge and personal insights to the product.

"He claims to have a deep comprehension of how to showcase artists to the public in a manner that made artists feel recognized and enthusiastic, and transformed casual listeners into devoted fans. This imbued the AI DJ with a sense of authenticity and credibility."

Jernigan states that the team aimed to leverage his knowledge of the music industry's business side, which he acquired through an MBA from Florida Agricultural and Mechanical University and a master's in music business from New York University.

He humorously claims that he is among the few individuals he knows who utilize all of their degrees on a daily basis.

Training the AI DJ

Spotify employs Sonantic, a cutting-edge text-to-speech AI voice technology that the company acquired in 2022, and OpenAI technologies to drive its AI DJ.

The AI DJ was trained using nearly 300 episodes of "The Get Up," with Jernigan's audio tracks isolated and fed to it to help analyze and understand his speech patterns, pitch, pacing, and emotions.

Jernigan maintained a record of the expressions he employs in daily discussions about music on his phone.

The AI DJ learned how to sound like him, mimicking his natural way of speaking and describing music, and then using that understanding to generate realistic-sounding audio outputs.

"This voice model is trained specifically on my voice, which is why it sounds so much like me," he explains. "Unlike other voice models, which train on a variety of voices and then add a layer on top, this one was trained solely on my voice, resulting in greater accuracy."

Jernigan's approach to training the AI DJ involves recording sessions, which he likens to voice acting. However, instead of reading lines and scripts as someone else, he reads them as himself. His goal is to sound as natural as possible during these sessions so that the AI model can learn to mimic his voice inflections.

"To ensure the voice model accurately pronounces words, he emphasizes the importance of being precise when hitting periods, pauses, and commas. He also stresses the need to enunciate T's correctly, especially if they appear at the end of words, as this can influence the model's pronunciation."

Jernigan regularly convenes with his international group of writers, music specialists, and data analysts to ensure the model remains current with emerging trends and music updates.

To keep up with the rapid-fire rap battle between rappers Kendrick Lamar and Drake, Jernigan's team updated the AI DJ with information about the songs and context about each artist.

When moments like that occur, Jernigan collaborates with his team of writers to promptly and precisely input the information into the AI DJ model to accurately mimic his voice.

"By having cultural music experts on the ground, we can inform the AI about current events, making the AI DJ experience feel more authentic."

Looking toward the future

The AI DJ feature of Spotify was initially launched in the U.S. and Canada in February, and is now accessible in 68 countries globally, including Australia, New Zealand, and certain European, Asian, Latin American, and African markets, according to the company.

Initially, Jernigan was nervous about how the world would receive him and the AI DJ, but his fear vanished once he saw how much Spotify users enjoyed the product.

""I have been recognized on the streets of London and Sydney, Australia, and it has been an honor to see the love and connection people have with the human behind the AI DJ," he says."

And Spotify is continuing to build upon its original AI DJ.

The company unveiled DJ Livi, a Spanish-speaking AI DJ that mimics the voice of Olivia Quiroz Roa, the company's senior music editor, on July 17.

In select markets, Spotify Premium users can now choose between listening to DJ Livi's music commentary in Spanish or DJ X's in English.

Spotify announced that introducing a Spanish-speaking DJ was a natural progression in the development of their product, and they are thrilled for the world to meet Livi.

As the product grows, Jernigan assures listeners that his voice and DJ X will remain a part of the show.

He says that he will continue to innovate and iterate, but he expects to be the voice for some time.

To stop worrying about money, enroll in CNBC's online course, Financial Wellness: Be Happier, Wealthier & More Financially Secure. We'll teach you the psychology of money, stress management, and healthy financial habits. Plus, use code EARLYBIRD for a 30% discount through September 2, 2024. Start today!

Sign up for CNBC Make It's newsletter to receive tips and tricks for success at work, with money and in life.

How I built my $400 million-a-year dating app Hinge
by Cheyenne DeVon

Make It