Ryu TAKEDA @ Spoken Dialogue System Group

Spoken Dialogue System

One of my goals is a maintainance-free spoken dialogue system that aquires neccesary knowledge during human interaction in the open world. My research interests are auditory scene analysis, knowledge acquisition during dialogue and a new application of audio processing techniques.

Research Area

Microphone Array Signal Processing
Sound Source Localization, Sound Source Separation, Echo Cancellation, Dereverberation
Language Processing
Language Modeling, Word Segmentation
Automatic Speech Recognition
Acoustic Modeling
Spoken Dialogue
Dialogue Management, Lexical Aquisition
Applications
Animal Speech Analysis (Frog, Insect)
The common techniuqes are statistical modeling and machine learning.

Main Works

Ryu Takeda and Kazunori Komatani: "Sound Source Localization based on Deep Neural Networks with Directional Activate Function Exploiting Phase Information," Proceedings of International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.405-409, 2016.
Ryu Takeda, Kazuhiro Nakadai and Kazunori Komatani: "Acoustic Model Training based on Node-wise Weight Boundary Model for Fast and Small-footprint Deep Neural Networks," Computer Speech & Language, Vol.46, pp.461-480, 2017.
Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Efficient Blind Dereverberation and Echo Cancellation based on Independent Component Analysis for Actual Acoustic Signals," Neural Computation, Vol.24, Issue 1, pp.234-272, 2012.
Kohei Ono, Ryu Takeda, Eric Nichols, Mikio Nakano and Kazunori Komatani: "Lexical Acquisition through Implicit Confirmations over Multiple Dialogues," Proceedings of Annual SIGdial Meeting on Discourse and Dialogue (SIGDIAL), pp.50-59, 2017.
Ikkyu Aihara, Ryu Takeda, Takeshi Mizumoto, Takuma Otsuka, Toru Takahashi, Hiroshi G. Okuno, Kazuyuki Aihara:"Complex and transitive synchronization in a frustrated system of calling frogs," Physical Review E, Vol.83, Issue 3.031913, 2011.

Ryu TAKEDA (Ph.D)

Affiliation
Department of Knowledge Science
The Institute of Scientific and Industrial Research (I.S.I.R.)
Osaka University
Address
Mihogaoka 8-1, Ibaraki, Osaka 567-0047, Japan
Contact
rtakeda [at] sanken.

Education

March 2006
Bachelor (Eng.), Faculty of Engineering, Undergraduate School of Informatics and Mathematical Science, Kyoto University, Japan

March 2008
Master (Informatics), Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University, Japan

March 2011
Ph.D. (Informatics), Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University, Japan

Professional Experience

April 2009 - March 2011
Research Fellow of the Japan Society for the Promotion of Science (DC2), Japan

April 2011 - September 2014
Researcher, Cental Research Laboratory, Hitachi, Ltd., Japan

October 2014 -
Research Associate, The Institute of Scientific and Industrial Research, Osaka University

September 2017 -
Visiting Researcher, The Language Technologies Institute, Carnegie Mellon University

Committee Services etc...

Memberships
The Institute of Electrical and Electronics Engineers (IEEE)
The Acoustical Society of Japan (ASJ)
The Information Processing Society of Japan (IPSJ)

Reviews
IEEE Transactions on Audio, Speech and Language Processing
Computer Speech & Language
IEICE Transactions on Information and Systems (Letter)
Journal of Robotics and Mechatronics
International Conference on Acoustics, Speech and Signal Processing (ICASSP 2009)
International Conference on Intelligent Robots and Systems (IROS 2011, 2015)
Annual Conference of the IEEE Industrial Electronics Society (IECON 2015)

Honors and Awards
Award for Entertainment Robots and Systems (NTF Award) Nomination Finalist (4/649), 22 Sep. 2008. "A Robot Listens to Music and Counts Its Beats Aloud by Separating Music from Counting Voice", IEEE/RSJ IROS-2008, Nice, Sep. 2008. (Takeshi Mizumoto, Ryu Takeda, Kazuyoshi Yoshii, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno)
Award for Entertainment Robots and Systems (NTF Award) Nomination Finalist (4/649), 22 Sep. 2008. "A Robot Uses Its Own Microphone to Synchronize Its Steps to Musical Beats While Scatting and Singing", IEEE/RSJ IROS-2008, Nice, Sep. 2008. (Kazumasa Murata, Kazuhiro Nakadai, Kazuyoshi Yoshii, Ryu Takeda, Toyotaka Torii, Hiroshi G. Okuno, Yuji Hasegawa, Hiroshi Tsujino)
RSJ/SICE AWARD for IROS2006 Best Paper Nomination Finalist, 2007/10/31. "Missing-Feature based Speech Recognition for Two Simultaneous Speech Signals Separated by ICA with a pair of Humanoid Ears", IEEE/RSJ IROS-2006, Beijing, China.
IEEE Robotics and Automation Society Japan Chapter Young Award, 2006/10/11. "Missing-Feature based Speech Recognition for Two Simultaneous Speech Signals Separated by ICA with a pair of Humanoid Ears", IROS-2006

Research Grants

Grant-in-Aid for Scientific Research (B) (Co-investigator), Apr. 2016 - Mar. 2020
Grant-in-Aid for Young Scientists (B) (Principal investigator), Apr. 2014 - Mar. 2016
Grant-in-Aid for JSPS Fellows (DC2), Apr.2009 - Mar. 2011

Skills

Programming: C/C++, octave (matlab)
Audio Signal Processing

Interests/Hobbies

Others
Turing as a runner: 2:46:03

Audio Signal Processing

Blind Dereverberation
Observation
(1st ch of 4ch)
Dereverbed
Blind Source Separation
Observation
(1st ch of 8ch)
Separated signal 1
Separated signal 2
The permutation is solved by using references to show the oracle performance.
Frog Speech Anaysis (Separation)
Observation (2nd ch of 3ch)
Frog A
Frog B
Frog C

Framework

HARK

For undergraduates

研究室に興味のある学生の方へ.

音声を使った簡単な実験はすぐに実行できます.
自分の音声を録音し,波形をプロットしてみましょう.
Samples にあるプログラムを実行し,出力された音声ファイルを聞いてみましょう.

フリーソフトで音声認識のカスタマイズも試せます.
Python 辺りに使い慣れておくと,音声認識結果などのテキスト処理をする上でも色々と楽です.

なお,プログラミングスキルは情報系で不可欠なので,色々とやってみることをお勧めします.
様々な経験(失敗)は早めにしておきましょう.
Softwares
  • vmware/virtual box: 仮想マシン上に Linux 環境を構築する際に必要.
  • wavesurfer/audacity : 音声の録音・スペクトルグラムの表示が可. マルチチャネルファイルも扱える.
  • Octave : 数値計算,グラフプロットなどが簡単. GUI も付属.
  • Python (numpy, scipy, matplotlib) : 同上.
  • OpenCV (C++, python): 画像処理用ライブラリ.手軽に動画・画像を扱える.
  • Julius: 音声認識ソフトウェア.Dictation kit を用いればすぐに音声認識を試せる.
References for training
  • MIT HRTF ( http://sound.media.mit.edu/resources/KEMAR.html ): 頭部伝達関数
  • 言語処理100本ノック ( http://www.cl.ecei.tohoku.ac.jp/nlp100/ ): 乾・岡崎研究室
  • フリーソフトでつくる音声認識システム
  • パターン認識と機械学習 上/下
  • 続・わかりやすいパターン認識

Samples

Released under the MIT license.
音声入力による単純な応答システム
予め登録している(単語, 応答文のペア)を用いて,発話に含まれる単語から応答文を選択して出力.
  • Procedure
  • Julius dictation kit を起動. (マイクは接続済みを仮定)
  • 音声認識結果を受け取り,認識単語から応答文を選択
  • 出力
  • Attention
  • Julius dictation kit は別途用意. Linux 上での実行がベスト.
  • 最小限の構成で作り込んではない.

sample code (python)
Independent Component Analysis for instantaneous mixture
瞬時混合モデルに基づく音源分離のサンプル
  • Procedure
  • 2つの音声信号を瞬時混合,観測音声の保存 (obs.wav)
  • 白色化とICA を適用し分離
  • 散布図のプロット, 分離音声の保存 (ica*.wav)

sample code (Octave)
Frequency-domain ICA for convolutive mixtures
周波数領域での瞬時混合モデルに基づく音源分離のサンプル
  • Procedure
  • 2つの音声信号を畳み込み混合, 観測音声の保存 (obs*.wav)
  • Frequency-domain ICA を適用し分離
  • Time-domain 瞬時混合 ICA を適用し分離
  • 散布図のプロット, 各種分離音声の保存 (fd-ica*.wav, td-ica*.wav)
  • Attention
  • パーミュテーションは解いていないので注意. 同梱データは解く必要なく揃う.
  • 帯域毎の分離誤差とパーミュテーション誤差は分けて考えること.

sample code (Octave)
Adaptive Filter
基本的な適応フィルタのサンプル
  • Procedure
  • 白色雑音,音声にインパルス応答を畳みこみ
  • LMS, NLMS, RLS, Kalman Filter でフィルタ推定
  • 誤差等のプロット

sample code (Octave)
Shared library
動的にリンクする共有ライブラリ実装のサンプル.
  • Procedure
  • dlopen, dlsym, dlclose を利用

sample code (C/C++)
Matrix vector multiplication using SSSE
行列・ベクトル積のSSSE命令による実装のサンプル.8-bit 16変数の積和を数命令で行う.
  • Attention
  • ベクトルの値は[0 1]に収まっていると仮定
  • メモリアライメントを揃える必要がある (posix_memalign)
  • Procedure
  • 行列の値を行で正規化し,8-bit signed char に変換
  • ベクトルの値を 8-bit unsigned char に変換
  • 16変数毎に ssse 命令で積和を計算・加算し,最後に結果のスケールを調整

sample code (C/C++)

Useful Tools

Kaldi
HTK
Open FST
Mecab
palmkit
yaml-cpp

Journal Papers

First
  1. Ryu Takeda, Kazuhiro Nakadai and Kazunori Komatani: "Acoustic Model Training based on Node-wise Weight Boundary Model for Fast and Small-footprint Deep Neural Networks," Computer Speech & Language, Vol.46, pp.461-480, 2017.
  2. Ryu Takeda and Kazunori Komatani:"Noise-robust MUSIC-based Sound Source Localization using Steering Vector Transformation for Small Humanoids," Journal of Robotics and Mechatronics, Vol.29, No.1, pp.26-36, 2017.
  3. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Efficient Blind Dereverberation and Echo Cancellation based on Independent Component Analysis for Actual Acoustic Signals," Neural Computation, Vol.24, Issue 1, pp.234-272, 2012.
  4. 武田 龍, 中臺 一博, 高橋 徹, 駒谷 和範, 尾形 哲也, 奥乃 博 :"残響下でのバージイン発話認識のための多入力独立成分分析を応用したロボット聴覚", 日本ロボット学会誌, Vol.27, No.7/8, pp.782-792, 2009.
  5. 武田 龍, 中臺 一博, 駒谷 和範, 尾形 哲也, 奥乃 博 : "独立成分分析に基づく適応フィルタのロボット聴覚への応用", 日本ロボット学会誌, Vol.26, No.6m pp.229-536, 2008.
Co-author
  1. Ikkyu Aihara, Ryu Takeda, Takeshi Mizumoto, Takuma Otsuka, Hiroshi G. Okuno:"Size Effect on Call Properties of Japanese Tree Frogs Revealed by Audio-Processing Technique," Journal of Robotics and Mechatronics, Vol.29, No.1, pp.247-254, 2017.
  2. Masahito Togami, Yohei Kawaguchi, Ryu Takeda, Yasunari Obuchi and Nobuo Nukaga: "Optimized speech dereverberation from probabilistic perspective for time varying acoustic transfer function," IEEE Transactions on Audio, Speech, and Language Processing, Vol.21, Issue 7, pp.1369-1380, 2013.
  3. Ikkyu Aihara, Ryu Takeda, Takeshi Mizumoto, Takuma Otsuka, Toru Takahashi, Hiroshi G. Okuno, Kazuyuki Aihara:"Complex and transitive synchronization in a frustrated system of calling frogs," Physical Review E, Vol.83, Issue 3. 031913 (2011) [5 pages], 21 Mar. 2011
  4. 駒谷 和範, 松山 匡子, 武田 龍, 高橋 徹, 尾形 哲也, 奥乃 博: "発語行為レベルの情報をユーザ発話の解釈に用いる音声対話システム", 情報処理学会論文誌, Vol.52, No.12,pp.3374-3385,2011.
  5. Takeshi Mizumoto, Ikkyu Aihara, Takuma Otsuka, Ryu Takeda, Kazuyuki Aihara, Hiroshi G. Okuno: "Sound Imaging of Nocturnal Animal Calls in Their Natural Habitat," Journal of Comparative Physiology A, Vol.197, No.9, pp.915-921, 2011. doi:10.1007/s00359-011-0652-7
  6. 村田 和真, 中臺 一博, 武田 龍, 奥乃 博, 長谷川 雄二, 辻野 広司: ロボットを対象としたビートトラッキングロボットの提案とその音楽ロボットへの応用", 日本ロボット学会誌, Vol.27, No.7, pp.793-801, 2009.
  7. 山本 俊一, 中臺 一博, 中野幹生, 辻野 広司, Jean-Marc Valin, 武田 龍, 駒谷 和範, 尾形 哲也, 奥乃 博: “遺伝的アルゴリズムを用いたパラメータ最適化による話者位置に基づく同時発話認識の向上”, ヒューマンインタフェース学会論文誌, Vol.8, No.2, pp.203-212, 2006.

Peer-Reviewed International Conference Papers

First
  1. Ryu Takeda and Kazunori Komatani: "Unsupervised Segmentation of Phoneme Sequences based on Pitman-Yor Semi-Markov Model using Phoneme Length Context," The 8th International Joint Conference on Natural Language Processing (IJCNLP), 2017. (accepted)
  2. Ryu Takeda, Kazuhiro Nakadai and Kazunori Komatani: "Node pruning based on Entropy of Weights and Node Activity for Small-footprint Acoustic Model based on Deep Neural Networks," Proceedings of INTERSPEECH, pp.1636-1640, Aug. 22, 2017. [52.3 % (799/1528)]
  3. Ryu Takeda and Kazunori Komatani: "Unsupervised Adaptation of Deep Neural Networks for Sound Source Localization using Entropy Minimization," Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.2217-2221, Mar. 7, 2017. [48.5% (1220/2518)]
  4. Ryu Takeda and Kazunori Komatani: "Bayesian Language Model based on Mixture of Segmental Contexts for Spontaneous Utterances with Unexpected Words," Proceedings of International Conference on Computational Linguistics (COLING), pp.161-170, Dec. 13, 2016. [32.4% (337/1039)]
  5. Ryu Takeda and Kazunori Komatani: "Discriminative Multiple Sound Source Localization based on Deep Neural Networks using Independent Location Model," Proceedings of IEEE Workshop on Spoken Language Technology (SLT), pp.603-609, Dec. 16, 2016. [60.9% (89/148):regular paper]
  6. Ryu Takeda and Kazunori Komatani: "Sound Source Localization based on Deep Neural Networks with Directional Activate Function Exploiting Phase Information," Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.405-409, Mar. 23, 2016. [47.1% (1265/2682)]
  7. Ryu Takeda, Kazuhiro Nakadai and Kazunori Komatani: "Acoustic Model Training based on Node-wise Weight Boundary Model Increasing Speed of Discrete Neural Networks," Proceedings of IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pp.52-58, Dec. 14, 2015. [47.8% (107/224)]
  8. Ryu Takeda and Kazunori Komatani: "Performance comparison of MUSIC-based sound localization methods on small humanoid under low SNR conditions," Proceedings of IEEE-RAS 15th International Conference on Humanoid Robots (Humanoids), pp.859--865, Nov. 4, 2015.
  9. Ryu Takeda, Naoyuki Kanda and Nobuo Nukaga: "Boundary contraction training for acoustic models based on discrete deep neural networks," Proceeding of INTERSPEECH, pp.1063-1067, 2014. [52.3% (614/1173)]
  10. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno : "Speedup and Performance Improvement of ICA-based Robot Audition by Parallel and Resampling-based Block-wise Processing", Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.1949-1956, IEEE, RSJ, Taipei, 18-22 Oct. 2010. [49.1%]
  11. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Upper-limit Evaluation of a Robot Audition based on ICA-BSS in Multi-source, Barge-in and Highly Reveberant Conditions," Proceedings of IEEE-RAS International Conference on Robotics and Automation (ICRA), pp.4366-4371, May 3-8, 2010, Anchorage, Aalaska, USA. [41.0% (847/2062)]
  12. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Automatic Estimation of Reverberation Time with Robot Speech to Improve ICA-based Robot Audition," Proceedings of IEEE-RAS Interanational Conference on Humanoid Robots (Humanoids), pp.250-355, IEEE, Paris, Dec. 7-10, 2009.
  13. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Step-size Parameter Adaptation of Multi-channel Semi-blind ICA with Piecewise Linear Model for Barge-in-able Robot Audition," Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp.2273-2282, IEEE, RSJ, St. Louis, 11-15 Oct. 2009. [54.5% (900/1650)]
  14. Ryu Takeda, Kazuhiro Nakadai, Toru Takahashi, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"ICA-based Efficient Blind Dereverberation and Echo Cancellation Method for Barge-in-able Robot Audition," Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp.3677-3680, Taipei, Taiwan, April 19 - April 24, 2009. [44.7% (1178/2633)] ((財)電気通信普及財団海外渡航助成)
  15. Ryu Takeda, Kazuhiro Nakadai, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Barge-in-able Robot Audition Based on ICA and Missing Feature Theory under Semi-Blind Situation," Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp.1718-1723, IEEE/RSJ, Nice, Sept. 2008.
  16. Ryu Takeda, Kazuhiro Nakadai, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno :"Exploiting Known Sound Sources to Improve ICA-based Robot Audition in Speech Separation and Recognition," Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp.1757--1762, San Diego, Oct. 2007.
  17. Ryu Takeda, Shun'ichi Yamamoto, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno : "Evaluation of Two Simultaneous Continous Speech Recognition with ICA BSS and MTF-based ASR," New Trends in Applied Artificial Intelligence (IEA/AIE), LNAI 4570, pp.384-394, Springer-Verlag. Kyoto, Jun. 2007
  18. Ryu Takeda, Shun'ichi Yamamoto, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno "Missing-Feature based Speech Recognition for Two Simultaneous Speech Signals Separated by ICA with a pair of Humanoid Ears," Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.878--885, Beijing, China, Sep. 2006 (IEEE Robotics and Automation Chapter, Japan 支部Young Award 受賞, RSJ/SICE AWARD for IROS2006 Best Paper Nomination Finalist)
  19. Ryu Takeda, Shun'ichi Yamamoto, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno "Improving Speech Recognition of Two simultaneous Speech Signals by Integrating ICA BSS and Automatic Missing Feature Mask Generation," Proceedings of International Conference on Spoken Language Processing (Interspeech), pp. 2302-2305, Pittsburgh, Sep. 2006. ((財) 原総合知的通信システム基金 国際会議論文発表助成).
Co-author
  1. Kohei Ono, Ryu Takeda, Eric Nichols, Mikio Nakano and Kazunori Komatani: "Lexical Acquisition through Implicit Confirmations over Multiple Dialogues," Proceedings of Annual SIGdial Meeting on Discourse and Dialogue (SIGDIAL), pp.50-59, Aug. 17, 2017. [40.9 % (47/115)]
  2. Kohei Ono, Ryu Takeda, Eric Nichols, Mikio Nakano and Kazunori Komatani: "Toward Lexical Acquisition during Dialogues through Implicit Confirmation for Closed-Domain Chatbots," Proceedings of Second Workshop on Chatbots and Conversational Agent Technologies (WOCHAT), 2016.
  3. Naoyuki Kanda, Ryu Takeda and Yasunari Obuchi: "Elastic spectral distortion for low resource speech recognition with deep neural networks," Proceedings of IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU), pp.309-314, 2013.
  4. Naoyuki Kanda, Ryu Takeda and Yasunari Obuchi: "Noise robust speaker verification with delta cepstrum normalization," Proceedings of INTERSPEECH, pp.3112-3116, 2013.
  5. Yasunari Obuchi, Ryu Takeda and Naoyuki Kanda: "Voice activity detection based on augmented statistical noise suppression, Proceedings of Asia-Pacific Signal & Information Processing Association Annual Summit and Conference (APISPA), pp.1-4, 2012.
  6. Masahito Togami, Yohei Kawaguchi, Ryu Takeda, Yasunari Obuchi and Nobuo Nukaga: "Multichannel speech dereverberation and separation with optimized combination of linear and non-linear filtering," Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.4057-4060, 2012.
  7. Naoyuki Kanda, Ryu Takeda, Yasunari Obuchi: "Using rhythmic features for Japanese spoken term detection," Proceedings of IEEE Spoken Language Technology Workshop (SLT), pp.170-175, 2012.
  8. Takeshi Mizumoto, Kazuhiro Nakadai, Takami Yoshida, Ryu Takeda, Takuma Otsuka, Toru Takahashi, Hiroshi G. Okuno: "Design and implementation of selectable sound separation on the Texai telepresence system using HARK," Proceedings of IEEE International Conference on Robotics and Automation (ICRA) , pp.2130--2137, 2011.
  9. Yasunari Obuchi, Ryu Takeda and Masahito Togami: "Bidirectional OM-LSA speech estimator for noise robust speech recognition," Proceedings of IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU), pp.173-178, 2011.
  10. Kyoko Matsuyama, Kazunori Komatani, Ryu Takeda, Toru Takahashi, Tetsuya Ogata, Hiroshi G. Okuno: "Analyzing User Utterances in Barge-in-able Spoken Dialogue System for Improving Identification Accuracy," Proceedings of International Conference on Spoken Language Processing (Interspeech), pp.3050--3053, 2010. (58.2%) Makuhari, 30 Sep.
  11. Ikkyu Aihara, Ryu Takeda, Takeshi Mizumoto, Takuma Otsuka, Toru Takahashi, Hiroshi G. Okuno: "Synchronization in Frustrated Calling Behavior of Japanese Tree Frogs," Conference on Dynamics in Systems Biology, Univ. of Aberdeen, UK, Sep. 16, 2009. (oral)
  12. Kazumasa Murata, Kazuhiro Nakadai, Kazuyoshi Yoshii, Ryu Takeda, Toyotaka Torii, Hiroshi G. Okuno, Yuji Hasegawa, Hiroshi Tsujino: "A Robot Uses Its Own Microphone to Synchronize Its Steps to Musical Beats While Scatting and Singing", Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.2459-, WeCT6.1, IEEE, RSJ, Nice, 24 Sep. 2008. (Award for Entertainment Robots and Systems (NTF Award) Nomination Finalist)
  13. Takeshi Mizumoto, Ryu Takeda, Kazuyoshi Yoshii, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno: "A Robot Listens to Music and Counts Its Beats Aloud by Separating Music from Counting Voice", Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.1538-1543, WeAT6.1 IEEE, RSJ, Nice, 24 Sep. 2008. (Award for Entertainment Robots and Systems (NTF Award) Nomination Finalist)
  14. Kazumasa Murata, Kazuhiro Nakadai, Kazuyoshi Yoshii, Ryu Takeda, Toyotake Torii, Hiroshi G. Okuno, Yuji Hasegawa, Hiroshi Tsujino: "A Robot Singer with Music Recognition Based on Real-Time Beat Tracking", Proceedings of 9th International Conference on Musical Information Retreival (ISMIR), pp.199-204, Philadelphia, 15 Sep. 2008.
  15. Kazumasa Murata, Kazuhiro Nakadai, Ryu Takeda, Hiroshi G. Okuno, Toyotaka Torii, Yuji Hasegawa, Hiroshi Tsujino: "A Beat-Tracking Robot for Human-Robot Interaction and Its Evaluation", Proceedings of IEEE-RAS Interanational Conference on Humanoid Robots (Humanoids), pp.79-84, Daejeon, Korea, Dec. 2, 2008.
  16. Shun'ichi Yamamoto, Ryu Takeda, Kazuhiro Nakadai, Mikio Nakano, Hiroshi Tsujino, Jean-Marc Valin, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno: “Recognition of Simultaneous Speech by Estimating Reliability of Separated Signals for Robot Audition”, PRICAI 2006: Trends in Artificial Intelligence, Lecture Note in Computatioal Science, No. 4099, pp.484-494, Springer-Verlag, Guilin, China, Aug. 2006
  17. Shun'ichi Yamamoto, Kazuhiro Nakadai, Mikio Nakano, Hiroshi Tsujino, Jean-Marc Valin, Ryu Takeda, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno: "Genetic Algorithm based Improvement of Robot's Hearing Capabilities in Separating and Recognizing Simultaneous Speech Signals", Advances in Applied Artificial Intelligence (IEA/AIE), Lecture Note in Artificial Intelligence, No. 4031, pp.207-217, Springer-Verlag. Annecy, France, Jun. 2006.
  18. Shun'ichi Yamamoto, Ryu Takeda, Kazuhiro Nakadai, Mikio Nakano, Hiroshi Tsujino, Jean-Marc Valin, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno: "Leak Energy based Missing Feature Mask Generation for ICA and GSS and Its Evaluation with Simultaneous Speech Recognition", Proceedings of ISCA Tutorial and Research Workshop on Statistical and Perceptual Audition (SAPA), pp.42-46, 2006

Review and Article

Co-author
  1. 山本 俊一, 武田 龍, 奥乃 博: "ミッシングフィーチャ理論に基づく音声認識を利用した複数話者同時発話認識", 計測と制御, Vol.46, No.6, pp.447-452, 2007.
  2. 合原 一究, 武田 龍, 水本 武志, 高橋 徹, 奥乃 博: "ニホンアマガエルの同期した発声行動に関する数理的研究および音響信号解析", 数理解析講究録, 1663, pp.153-158, 2009.

Domestic Meeting Papers

First
  1. 武田 龍, 中臺一博, 駒谷和範:"量子化 Deep Neural Network のための有界重みモデルに基づく音響モデル学習", 第46回 AIチャレンジ研究会, Nov. 2016.
  2. 武田 龍, 駒谷和範:"方向依存活性化関数を用いたDeep Neural Network に基づく識別的音源定位", 第112回音声言語情報処理研究会, July 2016.
  3. 武田 龍: "独立成分分析と雑音下音声認識技術によるロボット聴覚", 京都大学ICTイノベーション2009, 2009年2月20日
  4. 武田 龍,中臺 一博, 高橋 徹, 駒谷 和範, 尾形 哲也, 奥乃 博 :"独立成分分析を応用したロボット聴覚による残響下におけるバージイン発話認識", 日本ロボット学会第26回学術講演会, 1A2-02, Sept. 2008.
  5. 武田 龍, 中臺 一博, 駒谷 和範, 尾形 哲也, 奥乃 博:"ロボット音声対話のためのMFTとICAによるバージイン許容機能の評価", 情報処理学会第70回全国大会, 3U-1, Mar. 2008.
  6. 武田 龍, 中臺 一博, 駒谷 和範, 尾形 哲也, 奥乃 博 :"独立成分分析に基づく適応フィルタのロボット聴覚への応用", 日本ロボット学会第25回大会, 1N6 , Sept. 2007.
  7. 武田 龍, 山本 俊一, 駒谷 和範, 尾形 哲也, 奥乃 博 :"ICA と MFT に基づく音声認識における Soft Mask を用いた性能評価", 情報処理学会第69回全国大会, 6ZB-6, Mar. 2007.
  8. 武田 龍, 山本 俊一, 駒谷 和範, 尾形 哲也, 奥乃 博 :"ICAとミッシングフィーチャマスク自動生成によるロボット聴覚", 日本ロボット学会第24回大会, 1B13, Sept. 2006.
  9. 武田 龍, 山本 俊一, 駒谷 和範, 尾形 哲也, 奥乃 博 :"ICA による音源分離とミッシングフィーチャマスクの自動生成による同時発話認識", 情報処理学会第68回全国大会, March 2006.
Co-author
  1. 大野 航平, 武田 龍, エリック ニコルズ, 中野 幹生, 駒谷 和範, "対話を通じた未知語獲得に向けた暗黙的確認の提案", 第111回音声言語情報処理研究会, Mar. 2016.
  2. 大野 航平, 武田 龍, エリック ニコルズ, 中野 幹生, 駒谷 和範, "雑談対話における未知語や属性の獲得のための質問生成", 情報処理学会第78回全国大会, Mar. 2016.
  3. 梶野 尊弘, 武田 龍, 小路 悠介, 駒谷 和範, "マルチモーダル情報からの運転者の内部状態の推定", 情報処理学会第78回全国大会, Mar. 2016.
  4. 合原 一究, 水本 武志, 武田 龍, 大塚 琢馬, 高橋 徹, 奥乃 博, "アマガエルの多体系発声行動に潜む時空間ダイナミクス", 定量生物学の会・第2回年会, Jan. 2010
  5. 松山 匡子, 駒谷 和範, 武田 龍, 尾形 哲也, 奥乃 博: "バージイン発話タイミングを導入した指示対象同定", 情報処理学会音声言語研究会, May 2009. 学生奨励賞
  6. 合原 一究, 水本 武志, 武田 龍, 大塚 琢馬, 高橋 徹, 奥乃 博, "Time-delayed model on frogs' calling behavior and novel device for localizing calling animals," 第14回聴覚研究フォーラム, Mar. 2009.
  7. 合原 一究, 武田 龍, 水本 武志, 高橋徹, 合原一幸, 奥乃 博: "ニホンアマガエルの同期した発声行動に関する実験的研究およびその数理モデル解析," ニューロコンピューティング研究会, Mar. 2009.
  8. 大塚 琢馬, 村田 和真, 武田 龍, 中臺 一博, 高橋 徹, 尾形 哲也, 奥乃 博: "歌唱ロボットのためのビート情報とメロディ・ハーモニー情報の統合による音楽音響信号と楽譜の実時間同期手法の開発", 情報処理学会第71回全国大会, 5R-7, Mar. 2009.
  9. 松山 匡子, 駒谷 和範, 白松 俊, 武田 龍, 尾形 哲也, 奥乃 博: "実環境音声対話システムにおけるバージイン発話タイミングを活用した指示対象の同定", 情報処理学会第71回全国大会, 4Q-3, Mar. 2009
  10. 合原 一究, 武田 龍, 水本 武志, 高橋徹, 奥乃 博: "ニホンアマガエル3 匹の同期した発声行動に関する数理的・実験的研究," 第5回「生物数学の理論とその応用」, Jan. 2009.
  11. 合原 一究, 武田 龍, 水本 武志, 高橋徹, 奥乃 博: "アマガエルの同期した発声行動に関する実験的・数理的研究", (社)音響学会関西支部 第11回若手研究者交流研究発表会, Dec. 2008. 第11回若手研究者交流研究発表会 最優秀賞
  12. 村田 和真 中臺 一博, 武田 龍, 奥乃 博, 長谷川 雄二, 辻野 広司: "ビートトラッキングロボットの構築と評価", 第28回 AI チャレンジ研究会, SIG-Challenge-A802-3, pp.13--20, 人工知能学会, Nov. 2008. (社)人工知能学会 2008年度研究会優秀賞
  13. Ikkyu Aihara, Ryu Takeda, Toru Takahashi, Takeshi Mizumoto, Hiroshi G.Okuno: "Experimental and Theoretical Studies on Synchronized Calling Behavior of Three Japanese Tree Frogs", Japan-Slovenia Seminar on Nonlinear Science, Nov. 2008.
  14. 水本 武志, 武田 龍, 吉井 和佳, 高橋徹, 駒谷 和範, 尾形 哲也, 奥乃 博: "聴覚機能を持つ音楽ロボットのためのアーキテクチャの設計とビートカウントロボットへの適用", 26回ロボット学会学術講演会, 1A1-02, Sep. 2008.
  15. 水本 武志, 武田 龍, 吉井 和佳, 駒谷 和範, 尾形 哲也,奥乃 博: "音楽と自分の声を聞き分けながらビートに合わせて発声するロボットの開発", 情報処理学会第70回全国大会, 2X-8, Mar. 2008.

Patents

  1. 特願2014-204406(P2014-204406)・武本 剛, 本間 健, 武田 龍, 齋藤 仁, 畑山 正美, 三浦 春好, 天野 亘・設備点検支援装置・株式会社日立製作所, 東京都下水道サービス株式会社・2014年10月03日.
  2. 特願2014-192548(P2014-192548)・武田 龍, 本間 健, 武本 剛・音声認識方法、及び音声認識装置・株式会社日立製作所・2014年09月22日.
  3. 特願2014-190183(P2014-190183)・藤田 雄介, 武田 龍・検索サーバ、及び検索方法・株式会社日立製作所・2014年09月18日.
  4. 特願2013-178542(P2013-178542)・武田 龍・音声データ認識システム及び音声データ認識方法・株式会社日立製作所・2013年08月29日.
  5. 特願2010-124873(P2010-124873) 特許第5550456号(P5550456)・中臺 一博, 中島 弘史, 奥乃 博, 武田 龍・残響抑圧装置、及び残響抑圧方法・本田技研工業株式会社・2010年05月31日.
  6. 特願2010-105369(P2010-105369) 特許第5572445号(P5572445)・中臺 一博, 武田 龍, 奥乃 博・残響抑圧装置、及び残響抑圧方法・本田技研工業株式会社・2010年04月30日.
  7. 特願2009-166048(P2009-166048) 特許第5337608号(P5337608)・中臺 一博, 長谷川 雄二, 辻野 広司, 村田 和真, 武田 龍, 奥乃 博・ビートトラッキング装置、ビートトラッキング方法、記録媒体、ビートトラッキング用プログラム、及びロボット・本田技研工業株式会社・2009年07月14日.
  8. 特願2009-166049(P2009-166049) 特許第5150573号(P5150573)・中臺 一博, 長谷川 雄二, 辻野 広司, 村田 和真, 武田 龍, 奥乃 博・ロボット・本田技研工業株式会社・2009年07月14日.
  9. 特願2008-191382(P2008-191382) 特許第5178370号・武田 龍, 中臺 一博, 辻野 広司, 奥乃 博・音源分離システム・本田技研工業株式会社・2008年07月24日.