BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Vocal Tract Transfer Function Estimation Using Factor Analyzed \nT
 rajectory Hidden Markov Model - Tomoki Toda (Nara Institute of Science and
  Technology)
DTSTART:20080519T120000Z
DTEND:20080519T130000Z
UID:TALK12260@talks.cam.ac.uk
CONTACT:Dr Marcus Tomalin
DESCRIPTION:The estimation of the vocal tract transfer function (VTTF) for
  a speech signal \nis an essential problem in speech processing. Because t
 he speech signal \nresults from a convolution of the VTTF and a quasi-peri
 odic excitation \nsignal\, there are many missing frequency components bet
 ween adjacent \nharmonics of the fundamental frequency\, which make it ind
 eed hard to extract \nthe accurate VTTF. To address this problem\, I propo
 se a statistical approach \nto the offline VTTF estimation based on a fact
 or analyzed trajectory hidden \nMarkov model that effectively models harmo
 nic components observed over an \nutterance. This model is trained so that
  its likelihood for the observed \nharmonic component sequences is maximiz
 ed while considering VTTF parameters \nas hidden variables. The trained mo
 del enables the maximum a posteriori (MAP) \nestimation of a time-varying 
 VTTF sequence considering not only harmonic \ncomponents at each analyzed 
 frame but also those at other frames to \ninterpolate the missing frequenc
 y components in a probabilistic manner. The \neffectiveness of the propose
 d method is demonstrated by a result of a \nsimulation experiment.
LOCATION:LR5\, Engineering Department\, Baker Building
END:VEVENT
END:VCALENDAR