SUNGKYUNKWAN UNIVERSITY (SKKU), SEOUL, KOREA

Professor Ahn, Chang Wook (39) from the Computer Engineering Department has been developing an 'AI Composer' using deep-learning (a technique that makes the computer learn on its own) and an evolution algorithm (problem-solving process). The computer analyzes the inputs and combines them to a scale to produce thousands of songs and pick out the best one(s).

Professor Ahn has dreamed of developing the AI Composer since he was a graduate student back in early 2000s. As he started his Ph.D. program after finishing his master's degree, he changed his major from wireless communications to evolution computing in order to study artificial intelligence (AI). He said, "I had more fun studying AI than my original major and was interested in music since I was in high school. I even played acoustic guitar."

He is regularly having the AI Composer's songs evaluated by real composers. The professor said, "For now, AI is not able to create completely new or highly artistic songs like people do. However, composers are saying that its songs can be matched with that of amateur composers."

Ph.D. student, Chung, Jae Hoon has been developing the AI Composer along with Professor Ahn since the beginning. To fully understand the art of musical composition, he has been attending a music academy for two years now where he has been learning theory and how to generate compositions. Mr. Chung said, "I thought that to develop a humanlike AI Composer a profound understanding of music is required just as much as deep learning is."

Professor Ahn and Mr. Chung's study did not always go smoothly. They applied for research funding from many organizations, but got rejected. They are continuing their AI Composer study without any outside research funding. They cover necessary funds with profits from other research projects.

Professor Ahn's goal is to start a business with his students. He wants to develop a program that creates matching music to video, or matching video to music. The Professor says, "Now we are fusing music with AI only. We are trying to expand the area into various arts such as painting as well. We assume it will take about three to four years to develop these other fields."