Digital Signal Processor Lab (Funaki Lab.)

Research as a pillar of complex signal processing,Mobile phone, IP telephone voice recognition can be achieved in speech synthesis, speech enhancement,Conducting research on IP packet loss concealment methods required phone technology.

Complex speech analysis

Time-varying complex AR and has conducted its application in speech analysis method based on model building. The following analysis has been built.

  1. MMSE (EUSIPCO-98 @ Rhodes, Greece)
  2. Robust M estimation (ICSLP-98 @ Sydney)
  3. Auxiliary variable (IV) method (ICSLP-2000 @ Beijing)
  4. Generalized least squares (GLS)
  5. Expansion least-squares (ELS) (EUROSPEECH-2001 @ Aalborg, Denmark)
  6. Forward-backward LP-ELS (ICSLP-2002 @ Denver, USA)
  7. Forward-backward LP-GLS (MMSP-2002 @ St. Thomas, U.S. Virgin Islands)
  8. Output error by forward and backward LP-ELS (ISSPIT-2003 @ Darumushutaddo, Germany)

In addition, with the extension of the F0 estimation, speech enhancement, voice processing performs the evaluation and speech recognition.