Skip to content

Detail of publication


Hanzlíček, Z. and Matoušek, J. : First steps towards new Czech voice conversion system . Lecture Notes in Artificial Intelligence, 4188, p. 383-390, Springer, Berlin, 2006.


In this paper we deal with initial experiments on creating a new Czech voice conversion system. VC is a process which modifies the speech signal produced by source speaker so that it sounds like target speaker. Using VC technique a new voice for speech synthetizer can be gained with no need to record a huge amount of new speech data. The transformation is determined using equal sentences from both speakers; these sentences are time-aligned using modified DTW algorithm. The conversion is divided into two independent stages corresponding to the source-filter model of speech production. We employ conversion function based on GMM for transforming the spectral envelope described by LSFs. Residua are converted using so called residual prediction techniques. Unlike in other similar research works, we predict residua not from the transformed spectral envelope, but directly from the source speech. Four versions of residual prediction are described and compared in this study.

Detail of publication

Title: First steps towards new Czech voice conversion system
Author: Hanzlíček, Z. ; Matoušek, J.
Language: English
Date of publication: 11 Sep 2006
Year: 2006
Type of publication: Papers in journals
Title of journal or book: Lecture Notes in Artificial Intelligence
Series: 4188
Page: 383 - 390
ISBN: 3-540-39090-1
ISSN: 0302-9743
Publisher: Springer
Address: Berlin
Date: 11 Sep 2006 - 15 Sep 2006
/ 2013-01-25 11:10:20 /


voice conversion, speech synthesis, residual prediction


 author = {Hanzl\'{i}\v{c}ek, Z. and Matou\v{s}ek, J.},
 title = {First steps towards new Czech voice conversion system},
 year = {2006},
 publisher = {Springer},
 journal = {Lecture Notes in Artificial Intelligence},
 address = {Berlin},
 pages = {383-390},
 series = {4188},
 ISBN = {3-540-39090-1},
 ISSN = {0302-9743},
 url = {},