Flexible Speech Synthesis for Digital Media Production

2008 Fiscal Year Final Research Report

• Project/Area Number: 17300063
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Perception information processing/Intelligent robotics
• Research Institution: Nagoya Institute of Technology
• Principal Investigator: TOKUDA Keiichi Nagoya Institute of Technology, 大学院・工学研究科, 教授 (20217483)
• Co-Investigator(Kenkyū-buntansha): KITAMURA Tadashi 名古屋工業大学, 大学院・工学研究科, 教授 (60114865) ; NANKAKU Yoshihiko 名古屋工業大学, 大学院・工学研究科, 助教 (80397497) ; TODA Tomoki 端科学技術大学院大学, 情報科学研究科, 助教 (90403328)
• Co-Investigator(Renkei-kenkyūsha): BLACK Alan カーネギーメロン大学, 言語技術研究所, 准教授
• Project Period (FY): 2005 – 2008
• Keywords: 音声合成 / メディアコンテンツ / 声質 / 発話様式 / HMM

Research Abstract

本研究は, 「ト書き」付きの台本を声優の如く, 自在な表現で読み上げることのできる音声合成装置の実現を目指し, 種々の声質、スタイルの音声を少量の音声データから作成する"柔軟な" 音声合成技術を開発した。音声のみに止まらず、歌声についても研究対象とした.

Research Products

• [Journal Article] Tech-Ware : HMM-Based Speech Synthesis Resources (2009)
  • Author(s): Heiga Zen, Keiichi Tokuda
  • Journal Title: IEEE Signal Processing Magazine
• [Journal Article] Re-cent development of the HMM-based speech synthesis system (HTS) (2009)
  • Author(s): Heiga Zen, Keiichiro Oura, Takashi Nose, Junichi Yamagishi, Shinji Sako, Tomoki Toda, Takashi Masuko, Alan W. Black, Keiichi Tokuda
  • Journal Title: 2009 APSIPA Annual Summit and Con-ference, Sapporo Convention Center, Sapporo, Japan (to be published)
  • Peer Reviewed
• [Journal Article] Variational Bayesian method for HMM-based speech synthesis (2009)
  • Author(s): Kei Hashimoto, Heiga Zen, Yoshihiko Nankaku, Keiichi Tokuda
  • Journal Title: 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2009) Pages: 4029-4032
  • Peer Reviewed
• [Journal Article] Minimum generation error training by using original spectrum as reference for log spectral distortion measure (2009)
  • Author(s): Yi-Jian Wu, Keiichi Tokuda
  • Journal Title: 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2009) Pages: 4013-4016
  • Peer Reviewed
• [Journal Article] Voice conversion based on simultaneous modeling of spectrum and FO (2009)
  • Author(s): Kaori Yutani, Yosuke Uto, Yoshihiko Nankaku, Akinobu Lee, Keiichi Tokuda
  • Journal Title: 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2009) Pages: 3897-3900
  • Peer Reviewed
• [Journal Article] Statistical mapping between articulatory movements and acoustic spectrum with a Gaussian mixture model (2008)
  • Author(s): Tomoki Toda, Alan W. Black, and Keiichi Tokuda
  • Journal Title: Speech Communica-tion vol.50, no.3 Pages: 215-227
  • Peer Reviewed
• [Journal Article] Simultaneous Acoustic, Prosodic, and Phrasing Model Training for TTS Conversion Systems (2008)
  • Author(s): Keiichiro Oura, Yoshihiko Nankaku, Tomoki Toda, Keiichi Tokuda, Rannierry Maia, Shinsuke Sakai, Satoshi Nakamura
  • Journal Title: International Symposium on Chinese Spoken Language Processing (ISCSLP2008) Pages: 1-4
  • Peer Reviewed
• [Journal Article] Unsuper- vised Adaptation for HMM-Based Speech Synthesis (2008)
  • Author(s): Simon King, Keiichi Tokuda, Heiga Zen, Junichi Yamagishi
  • Journal Title: Interspeech 2008 Pages: 1869-1872
  • Peer Reviewed
• [Journal Article] Acoustic modeling with contextual additive structure for HMM-based speech recognition (2008)
  • Author(s): Yoshihiko Nankaku, Kazuhiro Naka- mura, Heiga Zen, Keiichi Tokuda
  • Journal Title: 2008 IEEE In- ternational Conference on Acous- tics, Speech, and Signal Processing (ICASSP 2008) Pages: 4469-4472
  • Peer Reviewed
• [Journal Article] Statistical approach to vocal tract transfer function estimation based on factor analyzed trajectory HMM (2008)
  • Author(s): Tomoki Toda, Keiichi Tokuda
  • Journal Title: 2008 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2008) Pages: 3925-3928
  • Peer Reviewed
• [Journal Article] ON the state definition for a trainable excitation model in HMM-based speech synthesis (2008)
  • Author(s): Ranniery Maia, Tomoki Toda, Kei- ichi Tokuda, Shinsuke Sakai, Satoshi Nakamura
  • Journal Title: 2008 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2008) Pages: 3965-3968
  • Peer Reviewed
• [Journal Article] Voice conversion based on maximum likelihood estimation of speech parameter trajectory (2007)
  • Author(s): Tomoki Toda, Alan W. Black, and Keiichi Tokuda
  • Journal Title: IEEE Transactions on Audio, Speech and Language Processing vol.15, no.8 Pages: 2222-2235
  • Peer Reviewed
• [Journal Article] Hidden Semi-Markov Model Based Speech Syn-thesis System (2007)
  • Author(s): Heiga Zen, Keiichi Tokuda, Takashi Masuko, Takao Kobayashi, and Tadashi Kitamura
  • Journal Title: IEICE Transactions on Information Systems vol.E90-D,no.5 Pages: 825-834
  • Peer Reviewed
• [Journal Article] Speech Parameter Generation Algo-rithm Considering Global Variance for HMM-Based Speech Synthesis (2007)
  • Author(s): Tomoki Toda and Keiichi Tokuda
  • Journal Title: IEICE Transactions on Information Systems vol.E90-D, no.5 Pages: 816-824
  • Peer Reviewed
• [Journal Article] Details of Nitech HMM-based speech synthesis system for the Blizzard Challenge 2005 (2007)
  • Author(s): Heiga Zen, Tomoki Toda, Masaru Nakamura, and Keiichi Tokuda
  • Journal Title: IEICE Transactions on Information and Systems vol.E90-D,no.1 Pages: 325-333
  • Peer Reviewed
• [Journal Article] Reformulating the HMM as a trajectory model by impos-ing explicit relationships between static and dynamic feature vector se-quences (2007)
  • Author(s): Heiga Zen, Keiichi Tokuda, Tadashi Kitamura
  • Journal Title: Computer Speech and Lan-guage vol.21, no.1 Pages: 153-173
  • Peer Reviewed
• [Journal Article] An HMM-based Brazilian Portuguese speech synthesizer and its characteristics (2006)
  • Author(s): Ranniery Maia, Heiga Zen, Keiichi Tokuda, Tadashi Kitamura, and Fer-nando G. V. Resende
  • Journal Title: Journal of Communication and Infor-mation Systems vol.21, no.2 Pages: 58-71
  • Peer Reviewed
• [Journal Article] 音声合成研究も協調と競争の時代にITheBliz-zard Challenges (2006)
  • Author(s): 徳田恵一, アランブラック
  • Journal Title: 日本音響学会誌 vol.62, no.6 Pages: 466-472
  • Peer Reviewed
• [Journal Article] Speaker adaptation of trajectory HMMs using feature-space MLLR (2006)
  • Author(s): Heiga Zen, Yoshihiko Nankaku, Keiichi Tokuda, Tadashi Kitamura
  • Journal Title: Interspeech 2006-ICLSP Pages: 1141-1144
  • Peer Reviewed
• [Journal Article] HMM-based singing voice synthesis system (2006)
  • Author(s): Keijiro Saino, Heiga Zen, Yoshihiko Nankaku, Akinobu Lee, Keiichi Tokuda
  • Journal Title: Interspeech 2006-ICSLP Pages: 2274-2277
  • Peer Reviewed
• [Journal Article] HMM-based European Portuguese TTS system (2005)
  • Author(s): Maria Joa o Barros, Ranniery Maia, Keiichi Tokuda, Fernando Gil Re- sende, Diamantino Freitas
  • Journal Title: INTERSPEECH 2005-EUROSPEECH Pages: 2581-2584
  • Peer Reviewed
• [Journal Article] A robust speaker-adaptive HMM-based text-to-speec hsynthesis
  • Author(s): Junichi Yamagishi, Takashi Nose, Heiga Zen, Zhen-Hua Ling, Tomoki Toda, Keiichi Tokuda, Simon King, Steve Renals
  • Journal Title: IEEE Transactions on Au-dio, Speech and Language Processing (accepted)
  • Peer Reviewed
• [Journal Article] Statistical parametric speech synthesis
  • Author(s): Heiga Zen, Keiichi Tokuda, Alan W. Black
  • Journal Title: Speech Communication (accepted)
  • Peer Reviewed
• [Journal Article] Spectral conversion based on statistical models including time-frequency matching
  • Author(s): Yoshihiko Nankaku, Kenichi Naka mura, and Keiichi Tokuda
  • Journal Title: Proc. of 6th ISCA Speech Synthesis Workshop CD-ROM proceedings)
  • Peer Reviewed
• [Presentation] HMMに基づく歌声合成のためのビブラートモデル化 (2009)
  • Author(s): 山田知彦, 武藤聡, 南角吉彦, 酒向慎司, 徳田恵一
  • Organizer: 情報処理学会研究報告「音楽情報科学(MUS)」
  • Place of Presentation: (vol2009-MUS-80,no.5,IPSJ-MUSO9080005)
  • Year and Date: 2009-05-14
• [Presentation] 声質変換における時系列マッチングを含む統計モデルの拡張 (2008)
  • Author(s): 油谷かおり, 南角吉彦, 戸田智基, 徳田恵一
  • Organizer: 日本音響学会2008年秋季研究発表会
  • Place of Presentation: 日本音響学会ボスター賞(vol.I,2-p-24, 411-412)
  • Year and Date: 20080900
• [Presentation] 因子分析に基づく固有声モデルを用いたHMM音声合成 (2007)
  • Author(s): 才野慶二郎, 全柄河, 南角吉彦, 李晃伸, 徳田恵一
  • Organizer: 日本音響学会2007年秋季研究発表会
  • Place of Presentation: (vol.I,3-4-8, 365-366)
  • Year and Date: 20070900
• [Presentation] HMM音声認識におけるコンテキストの加算的構造を考慮した音響モデリング (2007)
  • Author(s): 中村和寛, 全柄河, 南角吉彦, 李晃伸, 徳田恵一
  • Organizer: 日本音響学会2007年春季研究発表会
  • Place of Presentation: 日本音響学会ボスター賞(vol.I,1-P-13, 149-150)
  • Year and Date: 20070300
• [Presentation] Hidden Markov model-based speech synthesis as a tool for constructing comunicative spoken dialog systems (2006)
  • Author(s): Keiichi Tokuda
  • Organizer: Proc. of The 4th Joint Meeting of The Acoustical Society of America and The Acoustical Society of Japan
  • Place of Presentation: Honolulu, Hawaii
  • Year and Date: 20061128-122
• [Presentation] An HMM-based approach to flexible speech synthesis (2006)
  • Author(s): Keiichi Tokuda
  • Organizer: The 5th International Symposuim on Chinese Spoken Language Processing (ISCSLP 2006)
  • Place of Presentation: Kent Ridge, Singapore
  • Year and Date: 20061013-16
• [Presentation] トラジェクトリHMMの制約付き最尤線形回帰による話者適応 (2005)
  • Author(s): 全柄河, 南角吉彦, 徳田恵一, 村正
  • Organizer: 日本音響学会2005年秋季研究発表会講演論文集
  • Place of Presentation: 日本音響学会粟屋潔学術奨励賞(vol.I, 3-7-6, 113-114)
  • Year and Date: 20050900
• [Presentation] The trajectory HMM : Reformulating the HMM as a trajectory model (2005)
  • Author(s): Keiichi Tokuda
  • Organizer: Trajectory Models for Speech Processing, Supported by EPSRC (The Engineering and Physical Sciences Research Council, U.K.)
  • Place of Presentation: Edinburgh, U.K.
  • Year and Date: 2005-08-31
• [Book] "4.3HMM音声合成における統一的な韻律の制御"韻律と音声言語情報処理-アクセント・イントネーション・リズムの科学-(広瀬啓吉編著)(ISBN978-4-621-07674-3)
  • Author(s): 徳田恵一
  • Total Pages: 118-1279
  • Publisher: 丸善