Data-driven, nonlinear, formant-to-acoustic mapping for ASR

Philip Jackson; Boon Lo; Martin Russell

doi:10.1049/el:20020436

Data-driven, nonlinear, formant-to-acoustic mapping for ASR

Philip Jackson, Boon Lo, Martin Russell

Electronic, Electrical and Systems Engineering

Research output: Contribution to journal › Article

8 Citations (Scopus)

Abstract

With a view to using an articulatory representation in automatic recognition of conversational speech, two nonlinear methods for mapping from formants to short-term spectra were investigated: multilayered perceptrons (MLPs), and radial basis function (RBF) networks. Five schemes for dividing the TIMIT data according to their phone class were tested. The r.m.s. error of the RBF networks was 10%, less than that of the MLP, and the scheme based on discrete articulatory regions gave the greatest improvements over a single network.

Original language	English
Pages (from-to)	667-669
Number of pages	3
Journal	Electronics Letters
Volume	38
Issue number	13
DOIs	https://doi.org/10.1049/el:20020436
Publication status	Published - 1 Jan 2002

Access to Document

10.1049/el:20020436

Cite this

@article{c4e5983065fa4c95853b8009dc345270,

title = "Data-driven, nonlinear, formant-to-acoustic mapping for ASR",

abstract = "With a view to using an articulatory representation in automatic recognition of conversational speech, two nonlinear methods for mapping from formants to short-term spectra were investigated: multilayered perceptrons (MLPs), and radial basis function (RBF) networks. Five schemes for dividing the TIMIT data according to their phone class were tested. The r.m.s. error of the RBF networks was 10%, less than that of the MLP, and the scheme based on discrete articulatory regions gave the greatest improvements over a single network.",

author = "Philip Jackson and Boon Lo and Martin Russell",

year = "2002",

month = jan,

day = "1",

doi = "10.1049/el:20020436",

language = "English",

volume = "38",

pages = "667--669",

journal = "Electronics Letters",

publisher = "Institution of Engineering and Technology",

number = "13",

}

TY - JOUR

T1 - Data-driven, nonlinear, formant-to-acoustic mapping for ASR

AU - Jackson, Philip

AU - Lo, Boon

AU - Russell, Martin

PY - 2002/1/1

Y1 - 2002/1/1

N2 - With a view to using an articulatory representation in automatic recognition of conversational speech, two nonlinear methods for mapping from formants to short-term spectra were investigated: multilayered perceptrons (MLPs), and radial basis function (RBF) networks. Five schemes for dividing the TIMIT data according to their phone class were tested. The r.m.s. error of the RBF networks was 10%, less than that of the MLP, and the scheme based on discrete articulatory regions gave the greatest improvements over a single network.

AB - With a view to using an articulatory representation in automatic recognition of conversational speech, two nonlinear methods for mapping from formants to short-term spectra were investigated: multilayered perceptrons (MLPs), and radial basis function (RBF) networks. Five schemes for dividing the TIMIT data according to their phone class were tested. The r.m.s. error of the RBF networks was 10%, less than that of the MLP, and the scheme based on discrete articulatory regions gave the greatest improvements over a single network.

UR - http://www.scopus.com/inward/record.url?scp=0037142437&partnerID=8YFLogxK

U2 - 10.1049/el:20020436

DO - 10.1049/el:20020436

M3 - Article

VL - 38

SP - 667

EP - 669

JO - Electronics Letters

JF - Electronics Letters

IS - 13

ER -

Data-driven, nonlinear, formant-to-acoustic mapping for ASR

Abstract

Access to Document

Fingerprint

Cite this