Skip to content

Publications

Detail of publication

Citation

Vaněk Jan and Psutka Josef V. and Zelinka Jan and Trmal Jan : Training of Speaker-Clustered Discriminative Acoustic Models for Use in Real-Time Recognizers . Speech Processing, vol. 2010, p. 152-158, Institute of Photonics and Electronics AS CR, Prague, 2010.

Download PDF

PDF

Abstract

It is well known that gender-dependent (male/female) acoustic models are more acoustically homogeneous and therefore give better recognition performance than single gender-independent model in the case where the gender is successfully detected or a priory known. Speakers do not need to be split to two groups only. An algorithm to make higher number of speaker clusters is described in this paper. Further, the paper deals with a problem how to use these gender-based or speaker-clustered acoustic models in a real-time LVCSR where information from an automatic cluster detector is often delayed or incorrect. Moreover, various ways, how to incorporate discriminative training methods into training of the speaker-clustered acoustic models, are discussed in the paper.

Detail of publication

Title: Training of Speaker-Clustered Discriminative Acoustic Models for Use in Real-Time Recognizers
Author: Vaněk Jan ; Psutka Josef V. ; Zelinka Jan ; Trmal Jan
Language: English
Date of publication: 22 Sep 2010
Year: 2010
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: Speech Processing
Číslo vydání: 2010
Page: 152 - 158
ISBN: 978-80-86269-21-4
Publisher: Institute of Photonics and Electronics AS CR
Address: Prague
2011-03-15 16:21:58 / 2012-02-17 11:34:17 / 1

Keywords

Acoustics modeling, speaker-clustered models, automatic speech recognition

BibTeX

@INPROCEEDINGS{VanekJan_2010_Trainingof,
 author = {Van\v{e}k Jan and Psutka Josef V. and Zelinka Jan and Trmal Jan},
 title = {Training of Speaker-Clustered Discriminative Acoustic Models for Use in Real-Time Recognizers},
 year = {2010},
 publisher = {Institute of Photonics and Electronics AS CR},
 journal = {Speech Processing},
 address = {Prague},
 volume = {2010},
 pages = {152-158},
 ISBN = {978-80-86269-21-4},
 url = {http://www.kky.zcu.cz/en/publications/VanekJan_2010_Trainingof},
}