Skip to content

Detail of publication

Citation

Kala Jiří and Matoušek Jindřich and Tihelka Daniel : Reduction of Computing Cost in Unit Selection TTS System with Large Corpus . Speech Processing, p. 85-91, Institute of Photonics and Electronics AS CR, Praha, 2009.

Abstract

Unit selection concatenative speech synthesis requires high amount of computing time, especially for very large corpus. The corpus for our Czech text-to-speech system ARTIC contains over 650 thousands of diphone tokens. Unit selection system with such a large corpus is not well suitable for run-time speech generation when standard Viterbi search is used. In case of wide-range dialogue systems or other server applications, it is required to dispatch many requests at one moment and this could be problematic in relation to server resources. This paper describes various modifications of Viterbi search algorithm for massive reduction of computation cost in unit selection approach to speech synthesis.

Detail of publication

Title: Reduction of Computing Cost in Unit Selection TTS System with Large Corpus
Author: Kala Jiří ; Matoušek Jindřich ; Tihelka Daniel
Language: English
Year: 2009
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: Speech Processing
Page: 85 - 91
ISBN: 978-80-86269-18-4
Publisher: Institute of Photonics and Electronics AS CR
Address: Praha
Date: 1 Oct 2009
2011-03-15 16:21:44 / 2011-03-15 16:21:44 / 1

Keywords

text-to-speech, unit selection, large corpus, Viterbi algorithm

BibTeX

@INPROCEEDINGS{KalaJiri_2009_Reductionof,
 author = {Kala Ji\v{r}\'{i} and Matou\v{s}ek Jind\v{r}ich and Tihelka Daniel},
 title = {Reduction of Computing Cost in Unit Selection TTS System with Large Corpus},
 year = {2009},
 publisher = {Institute of Photonics and Electronics AS CR},
 journal = {Speech Processing},
 address = {Praha},
 pages = {85-91},
 ISBN = {978-80-86269-18-4},
 url = {http://www.kky.zcu.cz/en/publications/KalaJiri_2009_Reductionof},
}