UNOKI, Masashi Professor
School of Information Science, Human Life Design Area, School of Information Science
◆Degrees
M.S. and Ph.D. from Japan Advanced Institute of Science and Technology (1996,1999) 北陸先端科学技術大学院大学
M.S. and Ph.D. from Japan Advanced Institute of Science and Technology (1996,1999) 北陸先端科学技術大学院大学
◆Professional Experience
2000 - 2001 : 日本学術振興会特別研究員(DC2)(1998),ATR人間情報通信研究所 客員研究員(1999),ケンブリッジ大学CNBH客員研究員(2000-2001),日本学術振興会特別研究員(PD,北陸先端科学技術大学院大学 情報科学研究科)(1999-2001)
2000 - 2001 : JSPS Research Fellow (DC2)(1998), Visiting researcher, ATR Human Information Processing Laboratories (1999), Visiting Associate, CNBH, Univ. of Cambridge (2000-2001), JSPS Research Fellow (PD)(1999-2001)
◆Specialties
Intelligent robotics, Perceptual information processing, Intelligent informatics
◆Research Keywords
音声信号処理, 聴覚情景解析, 聴覚モデル, Speech dereverberation, Computational Auditory Scene Analysis, Auditory filterbank, Audio Information Hiding
◆Research Interests
Construction of the auditory filterbank
The aim of this work is to construct the auditory filterbank that can account psycho-acoustical data and physiological data for frequency selectivity of the auditory system.
Computational Auditory Scene Analysis
The study of computational theory of the auditory system tries to answer the following questions:
Extraction of the fundamental frequency of speech in real environments
Extraction of the fundamental frequency (F0) of target speech is an important problem not only in speech analysis/synthesis but also in various speech signal processings such as speech segregation. Various F0 estimation methods have been proposed, but the most of these methods have the drawbacks for estimating accurate F0s of target speech in real environments. My approach is to construct an estimation model based on computational auditory scene analysis (CASA).
A study on the speech dereverberation method
To dereverberate the original signal from a reverberant signal is an important issue concerning speech signal processing such as preprocessing for speech recognition systems. Most of the inverse filtering methods have to measure the impulse response of the room acoustics to determine its inverse filter before the dereverberation. Moreover, the impulse response temporally varies with various environmental factors (temperature etc.), so the room acoustics have to be measured each time these methods are used. In this work, it has been trying to model a speech dereverberation based on the Modulation Transfer Function, without measuring the impulse response of room acoustics.

■Publications

◆Published Papers
Semi-fragile speech watermarking based on singular-spectrum analysis with CNN-based parameter estimation for tampering detection
Galajit Kasorn, Karnjana Jessada, Unoki Masashi, Aimmanee Pakinee
APSIPA TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING, 8, -, 2019
Digital audio watermarking method based on singular spectrum analysis with automatic parameter estimation using a convolutional neural network
Kasorn Galajit, Jessada Karnjana, Pakinee Aimmanee, Masashi Unoki
Smart Innovation, Systems and Technologies, 110, 63-73, 2019
Enhanced feature network for monaural singing voice separation.
Weitao Yuan, Boxin He, Shengbei Wang, Jianming Wang,Array
Speech Communication, 106, 1-6, 2019
Estimates of Transmission Characteristics Related to Perception of Bone-Conducted Speech Using Real Utterances and Transcutaneous Vibration on Larynx.
Teruki Toya, Peter Birkholz, Masashi Unoki
Speech and Computer - 21st International Conference, SPECOM 2019, Istanbul, Turkey, August 20-25, 2019, Proceedings, 491-500, 2019
Data Augmentation for Monaural Singing Voice Separation Based on Variational Autoencoder-Generative Adversarial Network.
Boxin He, Shengbei Wang, Weitao Yuan, Jianming Wang, Masashi Unoki
IEEE International Conference on Multimedia and Expo, ICME 2019, Shanghai, China, July 8-12, 2019, 1354-1359, 2019
◆Misc
Study on modeling of room impulse response and its room acoustic characteristics
鵜木 祐史, 石川 大介, 柏原 佑太, 小林 まおり, 赤木 正人
電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 116, 302, 79-84, 2016
Estimation of Position Complexity based on Correlation between Evaluation Factors
竹内 章, 鵜木 祐史, 飯田 弘之
研究報告ゲーム情報学(GI), 2015, 13, 1-7, 2015
Study on blind method speech transmission index from noisy reverberant amplitude-modulated -signal
A. Miyazaki, S. Morita, M. Unoki
2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP’14), 105-108, 2014
A modulation-transfer-function-based method for restores sub-band power envelope from noisy reverberant speech
S. Morita, X. Lu, M. Unoki, M. Akagi, R. Hoffmann
The Acoustics 2012 Hong Kong Conference and exhibition, -, 2012
Study on Reversible Watermarking for Digital Audio Based on Cochlear Delay Characteristics
鵜木 祐史, 宮内 良太
電子情報通信学会技術研究報告 : 信学技報, 111, 287, 59-64, 2011
◆Books
音響情報ハイディング
コロナ社 ISBN:978-4-339-01135-7., 2018
「マスキング」,音響キーワードブック 日本音響学会編
コロナ社ISBN:978-4-339-00880-7, 2016
Method of Digital-Audio Watermarking Based on Cochlear Delay Characteristics, Multimedia Information Hiding Technologies and Methodologies for Controlling Data, Ed. Kazuhiko Kondo, Chapter 2
pp. 42-70, IGI Global, 2012
聴覚モデル,第5章 音の大きさのモデル(分担執筆)
129-167, コロナ社, 2011
Effects of spatial cues on detectability of alarm signals in noisy environments, PRINCIPLES AND APPLICATIONS OF SPATIAL HEARING, Edited by Yoiti Suzuki, Douglas Brungart, Hiroaki Kato, Kazuhiro Iida, Densil Cabrera, & Yukio Iwaya
484-493, World Scientific,, 2011
◆Conference Activities & Talks
Noise Suppression Method Based on Modulation Spectrum Analysis
20th International Conference on Speech and Computer SPECOM 2018 (SPECOM2018), Leipzig, Germany, 2018
Auditory-inspired end-to-end speech emotion recognition using 3D convolutional recurrent neural networks based on spectral-temporal modulation
IEEE International Conference on Multimedia and Expo (ICMC2018), San Diego, CA, USA, 2018
Speech watermarking based on robust principal component analysis and formant manipulations
2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP2018), Calgary, Canada, 2018
線形予測符号化方式におけるスペクトル拡散型音声電子透かしの検討
IEICE Technical Report, EMM2017-79, pp. 13-16, 奄美市名瀬公民館, 2018
雑音環境が駆動声の個人性・感情知覚に与える影響
日本音響学会聴覚研究会資料, vol. 48, no. 2, H-2018-35, pp. 175-180, 那覇, 2018

■Teaching Experience

Discrete Signal Processing, Human Perceptual Systems and its Models, Statistical Signal Processing, 離散信号処理特論, 知覚情報処理特論, 統計的信号処理特論

■Contributions to  Society

◆Academic Society Affiliations
信号処理学会, International Speech Communication Association, 電子情報通信学会, 日本音響学会, Research Institute of Signal Processing, Japan, Information and Communication Engineers, Institute of Electronics, The Acoustical Society of Japan, Institute of Electrical and Electronics Engineers, Acoustical Society of America
◆Academic Contribution
2017 International Workshop on Nonlinear Circuits and Signal Processing (NCSP17) Committee member (General Chair) , JAIST, Prof. Unoki Masashi , 2017 - 2017 , Guam, USA
2016 International Workshop on Nonlinear Circuits and Signal Processing (NCSP16) Committee member (General Vice Chair) , RISP , 2016 - 2016 , Honolulu, Hawaii, USA
The 14th IWDW, International Workshop on Digital-forensics and Watermarking (IWDW 2015) , Organizaing committee, , 2015 - 2015 , Tokyo University of Sciences

■Academic  Awards

・ Best paper award , Masashi Unoki , 11th International Conference on Social Computing and Social Media (SCSM 2019) , 2019
・ Best paper award , Masashi Unoki , 14th International Conference on Intelligent Information Hiding and Multimedia Signal Processing (II , 2018
・ 平成30年度支部学会活動貢献賞 , 日本音響学会北陸支部 , 2018