Application of Weighted Finite-State Transducers to Improve Recognition Accuracy for Dysarthric Speech

Omar Caballero Morales; Stephen Cox

Application of Weighted Finite-State Transducers to Improve Recognition Accuracy for Dysarthric Speech

Omar Caballero Morales, Stephen Cox

School of Computing Sciences

Research output: Contribution to conference › Paper

10 Citations (Scopus)

Abstract

Standard speaker adaptation algorithms perform poorly on dysarthric speech because of the limited phonemic repertoire of dysarthric speakers. In a previous paper, we proposed the use of "metamodels" to correct dysarthric speech. Here, we report on an improved technique that makes use of a cascade of Weighted Finite-State Transducers (WFSTs) at the confusion-matrix, word and language levels. This approach outperforms both standard MLLR and metamodels.

Original language	English
Pages	1761-1764
Number of pages	4
Publication status	Published - Sept 2008
Event	9th Annual Conference of the International Speech Communication Association (INTERSPEECH) - Brisbane, Australia Duration: 22 Sept 2008 → 26 Sept 2008

Conference

Conference	9th Annual Conference of the International Speech Communication Association (INTERSPEECH)
Country/Territory	Australia
City	Brisbane
Period	22/09/08 → 26/09/08

Cite this

@conference{ae9e4e8d0ad74840b4a1691d78da83c5,

title = "Application of Weighted Finite-State Transducers to Improve Recognition Accuracy for Dysarthric Speech",

abstract = "Standard speaker adaptation algorithms perform poorly on dysarthric speech because of the limited phonemic repertoire of dysarthric speakers. In a previous paper, we proposed the use of {"}metamodels{"} to correct dysarthric speech. Here, we report on an improved technique that makes use of a cascade of Weighted Finite-State Transducers (WFSTs) at the confusion-matrix, word and language levels. This approach outperforms both standard MLLR and metamodels.",

author = "{Caballero Morales}, Omar and Stephen Cox",

year = "2008",

month = sep,

language = "English",

pages = "1761--1764",

note = "9th Annual Conference of the International Speech Communication Association (INTERSPEECH) ; Conference date: 22-09-2008 Through 26-09-2008",

}

TY - CONF

T1 - Application of Weighted Finite-State Transducers to Improve Recognition Accuracy for Dysarthric Speech

AU - Caballero Morales, Omar

AU - Cox, Stephen

PY - 2008/9

Y1 - 2008/9

N2 - Standard speaker adaptation algorithms perform poorly on dysarthric speech because of the limited phonemic repertoire of dysarthric speakers. In a previous paper, we proposed the use of "metamodels" to correct dysarthric speech. Here, we report on an improved technique that makes use of a cascade of Weighted Finite-State Transducers (WFSTs) at the confusion-matrix, word and language levels. This approach outperforms both standard MLLR and metamodels.

AB - Standard speaker adaptation algorithms perform poorly on dysarthric speech because of the limited phonemic repertoire of dysarthric speakers. In a previous paper, we proposed the use of "metamodels" to correct dysarthric speech. Here, we report on an improved technique that makes use of a cascade of Weighted Finite-State Transducers (WFSTs) at the confusion-matrix, word and language levels. This approach outperforms both standard MLLR and metamodels.

M3 - Paper

SP - 1761

EP - 1764

T2 - 9th Annual Conference of the International Speech Communication Association (INTERSPEECH)

Y2 - 22 September 2008 through 26 September 2008

ER -