-  PhD University of Otago, Dunedin, New Zealand. 
      -  2002～ Senior Lecturer in the Department of Electrical and Computer Engineering/
                     The University of Auckland, New Zealand. 
      -  Visiting Researcher to Siena University (Italy), Essex University (UK), IDIAP    
         (Switzerland), Tokyo Institute of Technology (Japan), Electronic and  
         Telecommunication Research Institute (ETRI)(Korea) and many more.  
      - Collaborative work with Essex University, IDIAP, Hong Kong Polytechnic University,  
         Tokyo   Institute of Technology, and ETRI.        
        

7. 내용요약 :

  Humankind has dreamed of being able to verbally communicatingmachines in a way they used to be within themselves. That’s because speech is the most natural and easy way of communication and it is a rich information carrying signal. This signal carries information about speaker’s identity, emotion, age, gender, and many more. Our research focuses on proposing reliable vocal biometrics fromspeakers, which can be used effectively for identification with full-band and telephone-band speech in noisy environments. In this seminar we spotlight on mining speech signals to extract information about speakers’ identity. In speaker identification, a persontalks to a machine, then the machine identifies who has produced the speech signal. The process of identification comprises two stages; feature extraction and classification. In feature extraction, distinctive attributes from human voices are extracted to refer to the speakers. In classification, the extracted features are processed to find the identity of the most probable speaker. So far, not a single technique can be claimed to offer human-like identification performance in all environments (clean and noisy). Westrongly believe that the major problem is in the feature extraction stage rather than the classification stage.

       This talk is about shading light on a new method to extract a robust feature for text-independent  speaker identification systems in noisy environments. It is called perceptual log area ratio (PLAR).       
       The performances of this feature as compared to the commonly used features will be discussed. 
      Through this talk, we will try to put a convincing case to tell that there is still room to develop more  advanced features than the commonly used ones.