Newly developed software from McCombs researchers plays music according to listener’s mood

Mariane Gutierrez

After six years of work, researchers at the McCombs School of Business have developed a “personalized DJ” software that plays music according to a listener’s mood.

Researchers Maytal Saar-Tsechansky and Elad Liebman created an algorithm that identifies a user’s genre or song preferences and creates a music playlist according to the individual’s disposition.  The software is currently not open to the public, said Liebman, a senior data scientist at SparkCognition, an artifical intelligence program development company. 

“It initially doesn’t know much about you … (so) it tries out different types of songs, transitions, not just at random but in an intelligent way,” Liebman said. “It covers a lot of space, and it gets that feedback.”

The software began as a research project while Liebman was in graduate school. He said he was interested in combining his interests in music and research.

“I was interested in the ways that music preferences were manifesting themselves sequentially and how that was something that wasn’t sufficiently explored,” Liebman said.

The program works when a person gives the software their music preferences on a like or dislike system, Saar-Tsechansky said. The algorithm processes the user’s feedback to learn which song to play next.


“It’s a piece of software that interacts with listeners,” said Saar-Tsechansky, an information, risk, and operations management professor. “The listeners respond to the music they listen to, and it adapts what it plays over time.”

This algorithm is different compared to other music platforms like Spotify or Apple Music, Saar-Tsechansky said, because it focuses on the relevance of the experience, not just the playlists.

“A specific aspect that (the software) explores … is that we are trying to learn not just what kind of pieces of music the listener enjoys but also what sequence is pleasing and enjoyable. The relevance of DJs comes up here, where DJs don’t just play (a) random set of music,” Saar-Tsechansky said. “They try to build a sequence … so the sequence itself has value. It’s not a random sequence of a set of music they want to play.”

Management information systems senior Moises Gomez-Cortez said he could benefit from the software because he likes to code and listen to music.

“The research these professors are doing will benefit people by providing a better experience while listening to music,” Gomez-Cortez said. “It is also a great breakthrough for a software that can maybe rival big name music platforms.”