Journal "Software Engineering"
a journal on theoretical and applied science and technology
ISSN 2220-3397

Issue N6 2013 year

Semantic Network of Subject Area as a Basis for Transition Network Compilation for Automatic Recognition of Key Words in Speech Flow
A. A. Kharlamov , e-mail: kharlamov@analyst.ru, T. V. Ermolenko

The task of automatic speech recognition has not decided successfully now yet. But products such companies like Nuance [1], Autonomy [2], and others are known now. The products are used in applications, and in Russian language ones. Hidden Markov Models approach is good for the automatic speech recognition. Mistake of the recognition system in low room in such a case become less than 15 %. There is one, but essential lack: the transition network between word models compiled by hand. But the process is very labor consuming.

Keywords: speech recognition, key words, transition network between word models, text analysis, semantic network
pp. 33–40