Professor Chin-hui Lee from the School of Electrical and Computer Engineering of Georgia Institute of Technology was invited to visit the Key Laboratory of Speech Acoustic and Content Understanding of the Chinese Academy of Sciences on June 15, and to give a talk entitled “learning of deep architecture of acoustic and speech signal processing under the big data paradigm”.

In his lecture, Lee formulated a novel deep learning framework as finding a mapping function between an observed signal and a corresponding set of desired targets, in contrast to conventional theory-based signal processing.

Lee presented the recent research progress in his lab on speech enhancement, source separation, speech recognition, speech synthesis, voice conversion, voice activity detection and bandwidth expansion. They were all cast in a unified deep learning framework capable of jointly learning both some continuous speech parameters, such as log power spectrum in source separation, and some categorical information, such as voicing, phonation position and phone identity in speech synthesis.

It is believed that a combination of deep learning and big data offers vast opportunities for professionals to research and develop in the past, as well as in the future.

Professor Lee is a Fellow of IEEE and a Fellow of International Speech Communication Association (ISCA).

Professor Chin-hui Lee from the School of Electrical and Computer Engineering of Georgia Institute of Technology was giving his talk of “learning of deep architecture of acoustic and speech signal processing under the big data paradigm”. (Image by GE).

(Source: GE Fengpei from the Key Laboratory of Speech Acoustic and Content Understanding of the Chinese Academy of Sciences)