Fuzzy and speech recognition

Thomas Mantay (Thomas.Mantay@Informatik.Uni-Oldenburg.DE)
Fri, 8 Mar 1996 19:21:56 +0100


I'd like to develop an experimental speech recognition system that
uses fuzzy-clustering as classification component. To be more exact,
I want to partition the space of feature vectors by the fuzzy-c-means-
algorithm of Bezdek.

Concerning this I have some questions: Is it possible, say, for the
overaverage student, to develop and test such a system within a 3 months
period? Do I need tools like MatLab or the OGI tools etc. in order to
get along with the time restriction or is it still managable if I use
the C-sources that I find on the internet?

What do you think about fuzzy-algorithms in general in the field of
speech recognition? Is it worth studying the application of these
algorithms to ASR? Are there any researchers who tried to use fuzzy-algos
on ASR recently? The only researcher I found was Renato De Mori, but he
seems to have given up his approach.

I'd greatly appreciate if at least some of my questions would be answered.

Thanks in advance.

Best regards,

Thomas Mantay

*                                                                         *
* E-Mail:  Thomas.Mantay@arbi.Informatik.Uni-Oldenburg.DE                 *
* Real:    Thomas Mantay, Student of the University of Oldenburg, Germany *
*          Kaethe-Kollwitz-Str. 24                                        *
*          D-26127 Oldenburg                                              *
*          Germany                                                        *
*          Phone: 0441/3046501 (from within Germany)                      *
*                 +49/441/3046501 (from abroad)                           *
*                                                                         *