Artificial intelligence

Scientific journal

ISSN 2710-1673

ONLINE: ISSN 2710-1681

Search by:

Year of publication
Author name
Paper title

Speaker Adaptation Application in Isolated Word Recognition Systems

Seliukh R.1, Yukhymenko O.1
1 International Research and Training Centre for Information Technologies and Systems

Full text (PDF)

UDC: 004.934
Publication Language: Ukrainian
Stuc. intelekt. 2012; 17(3):185-193

Abstract: The paper deals with different variants of practical application of speaker adaptation for phoneme recognition of Ukrainian isolated words. The method of Maximum Likelihood Linear Regression (MLLR) is described. The matrixes of linear transformation are estimated in order to correct initial acoustic phoneme models. The results of experimental research of the adapted recognition system are discussed; particularly the amount of words in the adaptation sample is analyzed.

Keywords: speaker adaptation, maximum likelihood linear regression, linear transformation.

References:

View full text (PDF)