Summary: Modeling Consistency in a Speaker Independent
Continuous Speech Recognition System
Yochai Konig, Nelson Morgan, Chuck Wooters
International Computer Science Institute
1947 Center Street, Suite 600
Berkeley, CA 94704, USA.
Victor Abrash, Michael Cohen, Horacio Franco
333 Ravenswood Ave.
Menlo Park, CA 94025, USA
We would like to incorporate speakerdependent consistencies, such as
gender, in an otherwise speakerindependent speech recognition system.
In this paper we discuss a Gender Dependent Neural Network (GDNN)
which can be tuned for each gender, while sharing most of the speaker
independent parameters. We use a classification network to help generate
genderdependent phonetic probabilities for a statistical (HMM) recogni
tion system. The gender classification net predicts the gender with high
accuracy, 98.3% on a Resource Management test set. However, the in
tegration of the GDNN into our hybrid HMMneural network recognizer