Human Learning from Ears: Research on Otopecia

Period: April 2014 – March 2017
Grant category: Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research, Challenging Exploratory Research
Issue number: 26540092
URL: https://kaken.nii.ac.jp/ja/grant/KAKENHI-PROJECT-26540092/

To develop a learning tool that uses hearing, we have been researching speaker adaptation technology that converts speakers in speech synthesis, and have experimentally demonstrated that it is possible to digitally clone a voice from a small number of voice samples even when using deep learning instead of the hidden Markov model that has been used so far. We have also prototyped an audiobook reader iOS app that uses speaker adaptation technology in speech synthesis, and demonstrated that ePub format e-books can be read by a desired speaker. By constructing these experimental environments, we have been able to build a foundation for analyzing engagement time due to differences in speaker voices, and, in the future, to investigate the effects of differences in speaker voices on learning and memory.