This paper describes a technique for speaker-dependent wordspotting based on hidden Markov models (HMM's). The technique allows a speaker to specify keywords dynamically and to train the associated HMM's via a single repetition of a keyword. Non-keyword speech is modeled using an HMM trained from a prerecorded sample of continuous speech. The wordspotter is intended for interactive applications, such as the editing of voice mail or mixed-media documents, and for keyword indexing in single-speaker audio or video recordings. The performance of the system was tested on the Resource Management Database using 25 ship names as keywords. The probability of correct keyword detection was . 88 when the probability of a false alarm occurring in a sentence was . 03.
Bibliographic reference. Wilcox, Lynn D. / Bush, Marcia A. (1991): "HMM-based wordspotting for voice editing and indexing", In EUROSPEECH-1991, 25-28.