Hi everyone,

i'm doing a project which is 'speech recognition(of digits) using 
hybrid ANN(LVQ)-HMM'.. so far, i have an LPC cepstral coefficient from 
first time sampling : X[36frame x 12vector]. After the LPC cepstral 
coefficient has been weighted, it becomes Y[36frame x 12vector]. what 
should i do after i have this data? can i feed the data directly to 
LVQ?? if yes, how should i arrange them in order to be accepted by LVQ 
as input? should i combine some vary sampling data of a digit, or 
should i train them separately? i've read some about CMS, should i 
apply it to my system? please explain with detail formula, because i am 
new in this field. thank you so much.

P.S. : i'm using Matlab to this project

best regards, 

i-k

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