Professor Chang Joon-hyuk of the School of Electronic Engineering, together with KT, developed "Korean End-to-End Voice Recognition Transfer Learning" technology. The technology reduced data needed in learning while enhancing Korean voice recognition performance and proved its performance as a world-class voice recognition with a 13.7% decreased error rate compared to the previous best-performance deep learning voice recognition model. Also, he made the modification of voice recognition using a small amount of domain data by reutilizing previous models, not by starting from the beginning. KT, using this technology, said it is going to improve its AI service by using this technology.

This technology is made through a co-research from "AI One Team", an industry-university-research consultation of KT. Through this research, a total of four AI technologies were developed including Professor Chang's technology. The four technologies are △robot indoor space intelligence technology △ robot social interaction technology △ voice cloning technology △ Korean E2E voice recognition transfer learning technology, which will be used for KT AI businesses and services.

Meanwhile, the "AI One Team", which started out with a goal to make Korea the number one AI country, has a total of 10 industry-university cooperation associations, including KT, Hyundai Heavy Industries Group, LG Electronics, Korea Investment & Securities, Dongwon Group, Woori Bank, Hanjin, Korea Advanced Institute of Science and Technology (KAIST), Hanyang University, and Electronics and Telecommunications Research Institute(ETRI).