Plenary Lecture
Noise-Robust Speech Analysis
Professor Tetsuya Shimamura
Graduate School of Science and Engineering
Saitama University
Japan
E-mail: shima@sie.ics.saitama-u.ac.jp
Abstract: Digital signal processing and speech processing have evolved hand in hand over the last 50 years. Current speech processing techniques rely on advanced digital signal processing theories and algorithms. Especially, for speech analysis, this is true. In this plenary talk, speech analysis in noise, being a hot topic in speech processing, is discussed. Fundamental frequency (or pitch) estimation, formant frequency estimation and spectrum envelope estimation are considered. In a noise-free environment, these tasks are achieved by using autocorrelation function, cepstrum, linear prediction and so on. In a noisy environment, however, the performance of such speech analysis algorithms degrades suddenly. To solve the problem, new algorithms have been developed recent years. In this plenary talk, some of them are introduced. For fundamental frequency estimation, autocorrelation-based and modified autocorrelation-based approaches are shown. In such approaches, average magnitude difference function or cepstrum or both are effectively combined depending on the noisy situation. For formant and spectrum envelope estimation, noise compensation and pitch synchronous analysis approaches for linear prediction are shown. In such approaches, an iterative algorithm can be utilized to enhance the noise robustness. The performance of each algorithm will be demonstrated for synthetic and real speech data, in which a variety of noise conditions will be considered.
Brief Biography of the Speaker: Tetsuya Shimamura received the B.E., M.E., and Ph. D. degrees in electrical engineering from Keio University, Yokohama, Japan, in 1986, 1988, and 1991, respectively. In 1991, he joined Saitama University, Saitama City, Japan, where he is currently a Professor. During this, he joined Loughborough University, UK, and The Queen’s University of Belfast, UK, in 1995 and 1996, respectively, as a visiting Professor. He is an author or co-author of 6 books, and a member of the organizing committee of several international conferences. His interests are in digital signal processing and its applications to speech, image and communication systems. He received a Gold Paper Award at IEEE Pacific Rim Conference on Communications, Computers and Signal Processing in 2011. He also received a Best Paper Award from Research Institute of Signal Processing Japan in 2013.