Jason Pelecanos

Postdoctoral Researcher,
Conversational Biometrics Group

Human Language Technologies
IBM TJ Watson Research Labs

Biography:

Jason Pelecanos submitted his thesis in April 2003 for review as part of his PhD degree programme which began early 1998. He was granted the Australian Postgraduate Award (APA) Scholarship for funding his work at the Queensland University of Technology. The research topic was entitled "Robust Automatic Speaker Recognition". Prior to this effort, he earned the B.Eng Degree (majoring in Electrical and Electronic) with 1st class honours from the University of Southern Queensland. His undergraduate thesis project was entitled "Automatic Speaker Identification" for which he received a high distinction.

He is currently working as a Postdoctoral Researcher at the Human Language Technologies Group at the IBM T. J. Watson Research Center. His work focuses on core research in automatic speaker recognition technology and its commercial development. His research interests include improving and detecting speaker model robustness and enhancing the feature extraction process.

Selected Publications:

• J. Pelecanos, R. Vogt and S. Sridharan, “A Study on Standard and Iterative MAP Adaptation for Speaker Recognition”, in proc. of Australian International Conference on Speech Science and Technology, 2002.

• J. Pelecanos and S. Sridharan, “Feature Warping for Robust Speaker Verification”, in proc of A Speaker Odyssey, Paper 1038, 2001.

• I. McCowan, J. Pelecanos and S. Sridharan, “Robust Speaker Recognition using Microphone Arrays”, in proc. of A Speaker Odyssey, Paper 1024, 2001.

• N. Brummer and J. Pelecanos, “Unsupervised Evaluation of Speaker Verification Systems”, in proc. of A Speaker Odyssey, Paper 1027, 2001.

• J. Pelecanos and S. Sridharan, “Rapid Channel Compensation for One and Two Speaker Detection in the Nist 2000 Speaker Recognition Evaluation”, in Acoustics Australia, pp 17-20, Vol 29, No 1, 2001 (invited republication of above).

• J. Pelecanos, S. Myers and S. Sridharan, “Rapid Channel Compensation for One and Two Speaker Detection in the Nist 2000 Speaker Recognition Evaluation”, in proc. of Australian International Conference on Speech Science and Technology, pp 306-311, 2000.

• E. Wong, J. Pelecanos, S. Myers and S. Sridharan, “Language Identification using Efficient Gaussian Mixture Model Analysis”, in proc. of Australian International Conference on Speech Science and Technology, pp 78-83, 2000.

• S. Myers, J. Pelecanos and S. Sridharan, “Two Speaker Detection by Dual Gaussian Mixture Modeling”, in proc. of Australian International Conference on Speech Science and Technology, pp 300-305, 2000.

• J. Pelecanos, S. Myers, S. Sridharan and V. Chandran, “Vector Quantization based Gaussian Modelling for Speaker Verification”, in proc. of International Conference on Pattern Recognition, pp 298-301, Vol 3, 2000.

• J. Pelecanos, S. Slomka and S. Sridharan, “Enhancing Automatic Speaker Identification using Phoneme Clustering and Frame Based Parameter and Frame Size Selection”, in proc. of International Symposium on Signal Processing and its Applications, pp 633-636, Vol 2, 1999.

• J. Pelecanos and S. Sridharan, “Modelling Output Probability Distributions for Enhancing Speaker Recognition”, in proc. of Eurospeech, Vol 2, pp 999-1002, 1999.

• J. Moody, S. Slomka, J. Pelecanos and S. Sridharan, “On the Convergence of Gaussian Mixture Models: Improvements Through Vector Quantization”, in proc. of the International Conference on Spoken Language Processing, Paper 667, 1998.
  
  

  ---

  
  
  

  Contact Information:
  Jason Pelecanos
  Human Language Technologies
  Thomas J. Watson Research Center
  Rt. 134, Room 26-106
  Yorktown Heights, NY 10598 U.S.A.
  Tel.: (914) 945-2582
  Fax.: (914) 945-4490
  E-mail: Contact