Neural Process Lett
DOI 10.1007/s11063-013-9299-4
Extension of a Kernel-Based Classifier for Discriminative
Spoken Keyword Spotting
Shima Tabibian · Ahmad Akbari · Babak Nasersharif
© Springer Science+Business Media New York 2013
Abstract A keyword spotter is considered as a binary classifier that separates a class of utter-
ances containing a target keyword from utterances without the keyword. These two classes
are not inherently linearly separable. Thus, linear classifiers are not completely suitable for
such cases. In this paper, we extend a kernel-based classification approach to separate the
mentioned two non-linearly separable classes so that the area under the Receiver/Relative
Operating Characteristic (ROC) curve (the most common measure for keyword spotter evalu-
ation) is maximized. We evaluated the proposed keyword spotter under different experimental
conditions on TIMIT database. The results indicate that, in false alarm per keyword per hour
smaller than two, the true detection rate of the proposed kernel-based classification approach
is about 15 % greater than that of the linear classifiers exploited in previous researches. Addi-
tionally, area under the ROC curve (AUC) of the proposed method is 1% higher than AUC
of the linear classifiers that is significant due to confidence levels 80 and 95% obtained by
t -test and F-test evaluations, respectively. In addition, we evaluated the proposed method
in different noisy conditions. The results indicate that the proposed method show a good
robustness in noisy conditions.
Keywords Classifier · Discriminative keyword spotting · Kernel theory ·
Support vector machine
S. Tabibian (B )· A. Akbari · B. Nasersharif
Audio & Speech Processing Lab, Computer Engineering Department, Iran University
of Science & Technology, University St., Hengam Ave., Resalat Square, 16846-13114 Tehran, Iran
e-mail: shimatabibian@iust.ac.ir
URL: http://www.aspl.iust.ac.ir/
A. Akbari
e-mail: akbari@iust.ac.ir
B. Nasersharif
Electrical and Computer Engineering Department, K.N. Toosi University of Technology, Tehran, Iran
e-mail: bnasersharif@eetd.kntu.ac.ir
123