Title: On using warping function for LSFs transformation in a voice conversion system
Authors: Hanzlíček, Zdeněk
Matoušek, Jindřich
Citation: HANZLÍČEK, Zdeněk; MATOUŠEK, Jindřich. On using warping function for LSFs transformation in a voice conversion system. In: Proceedings of the 9th international conference on signal processing, ICSP '08, 26.10.2008 - 26.11.2008. 3rd ed. Beijing: IEEE Press, 2008, p. [1-4]. ISBN 978-1-4244-2178-7.
Issue Date: 2008
Publisher: IEEE Press
Document type: článek
URI: http://www.kky.zcu.cz/cs/publications/ZdenekHanzlicek_2008_OnUsingWarping
ISBN: 978-1-4244-2178-7
Keywords: hlasová konverze;deformační funkce
Keywords in different language: voice conversion;warping function
Abstract in different language: In this paper, a new approach to line spectral frequencies transformation is introduced and employed in the voice conversion framework. This approach stems from the fact that LSFs are some specific points on the frequency axis and their positions determine the shape of the spectral envelope. Thus, they could be transformed directly by frequency axis warping. Two warping functions were designed specially for LSFs and compared with the traditional GMM-based conversion function. Listening tests and mathematical evaluation revealed that speech transformed by using proposed warping functions is of higher quality and does not suffer from oversmoothing which is common for GMM-based transformation. On the other hand, the speaker identity is slightly better transformed by GMM-based conversion. However, it is possible to combine these two approaches to obtain a compromise between quality and speaker identity.
Rights: © Zdeněk Hanzlíček - Jindřich Matoušek
Appears in Collections:Články / Articles (KIV)

Files in This Item:
File Description SizeFormat 
ZdenekHanzlicek_2008_OnUsingWarping.pdfPlný text152,77 kBAdobe PDFView/Open
ZdenekHanzlicek_2008_OnUsingWarping_2.pdfPrezentace206,98 kBAdobe PDFView/Open

Please use this identifier to cite or link to this item: http://hdl.handle.net/11025/16948

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.