|
Showing 1 - 1 of
1 matches in All Departments
Automatic speech recognition (ASR) is a forefront of technology and
research today. The effectiveness of ASR depends upon the accurate
and quick classification of phonemes, which are the basic building
blocks of speech. To derive such a classifier for phoneme
classification in the context of ASR is the subject of my MASc
thesis at the University of Waterloo carried out in between April
2011 and July 2012 under the supervision of Professor Fakhreddine
Karray. Drawing upon several recent research topics applied to this
area, such as discriminative learning and locally adaptive metrics,
a novel classifier referred to as the discriminative
locally-adaptive nearest centroid classifier (DLANC). DLANC is
structurally simple, very quick to train on even very large sets of
data, and it also produces very good classification results on
standard TIMIT data. This book describes the DLANC classifier in
detail, including its background and how it is derived. A detailed
comparison between the DLANC classifier and several other existing
classifiers for phoneme classification are made on standard TIMIT
data. Numerous illustrations and diagrams make many theoretical
points easy to understand.
|
You may like...
The Wonder Of You
Elvis Presley, Royal Philharmonic Orchestra
CD
R71
R60
Discovery Miles 600
Loot
Nadine Gordimer
Paperback
(2)
R398
R330
Discovery Miles 3 300
|
Email address subscribed successfully.
A activation email has been sent to you.
Please click the link in that email to activate your subscription.