V. Smirnov, M. Gusev, M. P. Farkhadov, “Acoustic modeling module function in a system for automatic analysis of unstructured speech data”, UBS, 2013, Issue 45,Pages <nobr>181

Information Technology Applications in Control

Acoustic modeling module function in a system for automatic analysis of unstructured speech data

V. Smirnov^a, M. Gusev^b, M. P. Farkhadov^c

^a Speech Drive, ltd. – company
^b Research Institute “KVANT”
^c Institute of Control Sciences of RAS

Abstract: We focus on acoustic modeling – a component of a system for automatic analysis of unstructured speech data, which aims at creating statistical description for speech sounds. In the first part of the paper we outline goals of a system for automatic analysis of unstructured speech data and explain the place and functions of acoustic modeling within the system. Then we suggest an algorithm of acoustic modeling and address each of its stages in detail.

Keywords: acoustic modeling, hidden Markov models, forward-backward algorithm, Baum-Welch re-estimation, classification and regression trees, automatic analysis of unstructured speech data.

UDC: 004.52 + 004.934
BBK: 22.1