Hi All.
I'm bothered by the following in the process of
implementing a speaker independent ASR system. Say i
have say 6 word utterances(6 speakers) per word to be
recognized. I wish to compute the distance between an
incoming speech sample and these 12-coefficient
template vectors. How will i go about computing the
distances, since speech is going to be more than one
frame, in fact several frames in duration and
considering the fact that these utterances are likely
to be of unequal duration. Am afraid i have not been
able to get this info from available documentation, as
everybody talks as if you only need to use a single
feature vector( that is for one frame only), for the
necessary computations and decision. Kindly help me
resolve this confusion.
Thanks

______________________________
New Year Gift for Members of DSPRelated.com.  Details here.

(You need to be a member of speech-recognition -- send a blank email to speech-recognition-subscribe@yahoogroups.com )