Dynamic classifier combination in hybrid speech recognition systems using utterance-level confidence values

Dynamic classifier combination in hybrid speech recognition systems using utterance-level confidence values (1999)

Cached

Download Links

by Katrin Kirchhoff , Ag Angewandte Informatik , Technische Fakultät

Venue:	Proceedings ICASSP-99
Citations:	20 - 3 self

BibTeX

@INPROCEEDINGS{Kirchhoff99dynamicclassifier,
    author = {Katrin Kirchhoff and Ag Angewandte Informatik and Technische Fakultät},
    title = {Dynamic classifier combination in hybrid speech recognition systems using utterance-level confidence values},
    booktitle = {Proceedings ICASSP-99},
    year = {1999},
    pages = {693--696}
}

OpenURL

Abstract

A recent development in the hybrid HMM/ANN speech recognition paradigm is the use of several subword classifiers, each of which provides different information about the speech signal. Although the combining methods have obtained promising results, the strategies so far proposed have been relatively simple. In most cases frame-level subword unit probabilities are combined using an unweighted product or sum rule. In this paper, we argue and empirically demonstrate that the classifier combination approach can benefit from a dynamically weighted combination rule, where the weights are derived from higher-than-frame-level confidence values. 1.

Keyphrases

utterance-level confidence value hybrid speech recognition system sum rule combination approach recent development case frame-level subword unit probability unweighted product higher-than-frame-level confidence value several subword classifier different information weighted combination rule speech signal promising result