Back to Search

The effects of stress and pronunciation modeling errors in the SMM-based Turkish MKS system.

Title	The effects of stress and pronunciation modeling errors in the SMM-based Turkish MKS system.
Author	Güner, Ekrem, Demiroğlu, Cenk
Publication Date:	2011
Publication Place	- IEEE
Subject	Hidden Markov models, Speech synthesis
Type	Document
Language	Turkish
Digital	Yes
Manuscript	No
Library:	Özyeğin University
Library Asset ID	978-1-4577-0462-8
Record ID	84ef9fb6-b7b5-41cc-814b-d54d4919df49
Library Location	Electrical & Electronics Engineering
Date	2011
Notes	Due to copyright restrictions, the access to the full text of this article is only available via subscription.
Sample Text	SMM-based text-to-speech synthesis (SMM-MKS) method attracts the attention of more and more researchers day by day. One of the most important advantages of this method is the absence of distortion effects seen in unit selective systems. In this paper, the performance of the first academic Turkish SMM-MKS system is reported. Although Turkish is a language that is read as it is written, this conversion is not always one-to-one. Moreover, although stress in Turkish depends on certain rules, it may not always be possible to use them for accurate stress marking. Therefore, in addition to the quality of our basic design system, its sensitivity to pronunciation and accentuation errors is also examined and presented. We observed that using a complex pronunciation and stress model affects the duration of phonemes the most, which increases the quality, but when they are used together, their contributions are not superimposed.
DOI	10.1109/SIU.2011.5929807

View in source Özyeğin University Özyeğin University - Ottoman library catalog search

Özyeğin University

The effects of stress and pronunciation modeling errors in the SMM-based Turkish MKS system.

Author Güner, Ekrem, Demiroğlu, Cenk

Publication Date 2011

Publication Place - IEEE

Subject Hidden Markov models, Speech synthesis

Type Document

Language Turkish

Digital Yes

Manuscript No

Library Özyeğin University

Library Asset ID 978-1-4577-0462-8

Record ID 84ef9fb6-b7b5-41cc-814b-d54d4919df49

Library Location Electrical & Electronics Engineering

Date 2011

Notes Due to copyright restrictions, the access to the full text of this article is only available via subscription.

Sample Text SMM-based text-to-speech synthesis (SMM-MKS) method attracts the attention of more and more researchers day by day. One of the most important advantages of this method is the absence of distortion effects seen in unit selective systems. In this paper, the performance of the first academic Turkish SMM-MKS system is reported. Although Turkish is a language that is read as it is written, this conversion is not always one-to-one. Moreover, although stress in Turkish depends on certain rules, it may not always be possible to use them for accurate stress marking. Therefore, in addition to the quality of our basic design system, its sensitivity to pronunciation and accentuation errors is also examined and presented. We observed that using a complex pronunciation and stress model affects the duration of phonemes the most, which increases the quality, but when they are used together, their contributions are not superimposed.

DOI 10.1109/SIU.2011.5929807