SHENG LI (李 勝) Japanese version

Ph.D.
Researcher
Kyoto University, School of Informatics
Department of Intelligence Science and Technology
Speech and Audio Processing Lab.

RESEARCH FOCUS:

• Acoustic modeling for LVCSR systems (English, Chinese, Japanese)
  familiar with Cambridge GALE Mandarin system, CAS Acoustic Institution Mandarin system, Kyoto University Japanese CSJ system, NICT English spoken lecture system

• ASR result confidence estimation and correction.

• Lightly-supervised/semi-supervised/ensemble acoustic model training (spoken lectures).

• Acoustic model and language model adaptation (spoken lectures).

EDUCATION:

• 2006.7 B.S Computer Science,
  Nanjing University

• 2009.7 M.E Software and Embedded Systems,
  Nanjing University（Joint Program of Chinese Academy of Sciences, Chinese University of Hong Kong and Nanjing University）

• 2016.3 Ph.D Informatic Science,
  Kyoto University (supported by Japanese Government)

PUBLICATIONS:

Ph.D Thesis:

• Sheng Li (supervised by Prof. Tatsuya Kawahara).
  Speech Recognition Enhanced by Lightly-supervised and Semi-supervised Acoustic Model Training.
  Ph.D. Thesis, Kyoto University, Feb, 2016.

Journals:

• S.Li, Y.Akita, and T.Kawahara.
  Semi-supervised acoustic model training by discriminative data selection from multiple ASR systems' hypotheses.
  IEEE Trans. Audio, Speech \& Language Process., Vol.24, No.9, pp.1520--1530, 2016. (text) (PDF)

• S.Li, Y.Akita, and T.Kawahara.
  Automatic lecture transcription based on discriminative data selection for lightly supervised acoustic model training.
  IEICE Trans., Vol.E98-D, No.8, pp.1545--1552, 2015. (text) (PDF)

• L. Wang, H. Chen, S. Li and H. Meng.
  Phoneme-level articulatory animation in pronunciation training,
  Speech Communication, Vol. 54, Issue 7, Sept. pp. 845–856, 2012. (text) (PDF)

International Conferences:

• S.Li, X.Lu, S.Sakai, M.Mimura and T.Kawahara
  SEMI-SUPERVISED ENSEMBLE DNN ACOUSTIC MODEL TRAINING
  Accepted in IEEE-ICASSP, 2017.

• S.Li, X.Lu, S.Mori, Y.Akita, T.Kawahara.
  Confidence Estimation for Speech Recognition Systems using Conditional Random Fields Trained with Partially Annotated Data
  Int'l Sympo. Chinese Spoken Language Processing (ISCSLP), 2016. (PDF)

• S.Li, Y.Akita, and T.Kawahara.
  Data selection from multiple ASR systems' hypotheses for unsupervised acoustic model training.
  In Proc. IEEE-ICASSP, pp.5875--5879, 2016. (text) (PDF)

• S.Li, Y.Akita, and T.Kawahara.
  Discriminative data selection for lightly supervised training of acoustic model using closed caption texts.
  In Proc. INTERSPEECH, pp.3526--3530, 2015.(oral) (text) (PDF)

• S.Li, X.Lu, Y.Akita, and T.Kawahara.
  Ensemble speaker modeling using speaker adaptive training deep neural network for speaker adaptation.
  In Proc. INTERSPEECH, pp.2892--2896, 2015. (text) (PDF)

• S.Li, Y.Akita, and T.Kawahara.
  Corpus and transcription system of Chinese Lecture Room.
  In Proc. Int'l Sympo. Chinese Spoken Language Processing (ISCSLP), pp.442--445, 2014. (text) (PDF)

• S. Li and L. Wang.
  Cross Linguistic Comparison of Mandarin and English EMA Articulatory Data,
  In Proc. INTERSPEECH, 2012. (Travel granted by IBM research) (text) (PDF)

• S. Li, L. Wang and E. Qi.
  The Phoneme-level Articulator Dynamics for Pronunciation Animation,
  In Proc. IALP, Nov.15-17, Pages 283-286, 2011. (text) (PDF)

• J. Chen, L. Wang, C. Li, J. Hu and S. Li.
  IELS: A Computer-aided Pronunciation Training System for Undergraduate Students,
  ICETC, Vol.1, pp.338-342, 2010. (text) (PDF)

Technical Reports:

• S. Li, X. Lu, S. Sakai, T. Kawahara,
  Diversity-driven Semi-supervised Ensemble DNN Acoustic Model Training,
  ASJ autumn, 2016.

• S. Li, X. Lu, S. Sakai, T. Kawahara,
  Diversity-driven Semi-supervised Ensemble DNN Acoustic Model Training,
  IPSJ IEICE-SP2016-40, 2016. (oral)

• S. Li, Y. Akita, T. Kawahara,
  Discriminative data selection from multiple ASR systems' hypotheses for unsupervised acoustic model training,
  IPSJ SIG-SLP-109-8, 2015.(oral)

• S. Li, Y. Akita, T. Kawahara,
  Effective combination of multiple ASR hypotheses with CRF-based classifiers,
  ASJ autumn, 2015.

• S. Li, Y. Akita, T. Kawahara,
  Incorporating divergences from hypotheses of multiple ASR systems to improve unsupervised acoustic model training,
  ASJ spring, 2015.

• S. Li, Y. Akita, T. Kawahara,
  Unsupervised Training of Deep Neural Network Acoustic Models for Lecture Transcriptions,
  ASJ autumn, 2014.

• S. Li, Y. Akita, T. Kawahara,
  Classifier-based data selection for lightly-supervised training of acoustic model for lecture transcription,
  IPSJ SIG-SLP-102-4, 2014.(oral)

• S. Li, Y. Akita, T. Kawahara,
  Data Selection Assisted by Caption to Improve Acoustic Modeling for Lecture Transcription,
  ASJ spring, 2014.(oral)

• S. Li, M. Mimura, T. Kawahara,
  Automatic Transcription of Chinese Spoken Lectures,
  ASJ autumn, 2013.

• S. Li, K. Luo and L. Wang,
  The Phoneme-level Articulator Dynamics for 3D Pronunciation Animation for Chinese,
  Bulletin of Advanced Technology Research, Vol.5 No.10/Otc.2011, Pages 5-7.

• S. Li and C. Li,
  Application of the RFID based audio service in regional navigation system,
  Bulletin of Advanced Technology Research, Vol.3 No.2/Feb.2009, Pages 44-47.

Research Projects:

AWARDS:

ACADEMIC SERVICES:

MEMBERSHIP:

SOFTWARE & RECIPES:

DATA RELEASES:

ALUMNI:

Last update: 2016-8-20