National Technical University of Athens (NTUA)
Βελτίωση της ποιότητας συνθετικής φωνής και εφαρμογή σε σύγχρονα τηλεπικοινωνιακά περιβάλλοντα και υπηρεσίες
Abstract
dc:descriptionThe subject of this thesis is speech synthesis technology and, in particular, the improvement of the quality of Text‐to‐Speech (TTS) systems for application in contemporary telecommunication environments and services. Emphasis is given in Corpus‐based Speech Synthesis and in Unit Selection Concatenative TTS systems by focusing on the Unit Selection algorithm and the design of the cost functions which comprising it. Methods and approaches concerning the implementation of not only General Domain TTS systems, but also adapted scaled‐down TTS systems for computational environments with limited resources and embedded systems in general, are explored and evaluated. In addition, contemporary Statistical Parametric Speech Synthesis based on Hidden Markov Models is explored for the case of the Greek language. More particularly, this thesis deals with research efforts and contributes to the following: - The design and implementation of the unit selection algorithm for a general purpose Textto‐Speech system for the Greek language. - The design and implementation approaches for the efficient integration of Unit Selection technology in computational environments with limited resources, such as mobile devices, with no considerable speech quality degradation. In particular, the issues of database reduction, acoustic inventory compression and runtime computational load minimization are mainly addressed. Both objective and subjective assessments confirm the effectiveness of these approaches in terms of constructing a general purpose embedded unit selection TTS system and reducing the computational requirements while maintaining high speech quality. - The introduction of one‐class classification as a framework for the spectral join cost calculation in unit selection speech synthesis. A data‐driven approach is adopted which exploits the natural similarity of consecutive speech frames in the speech database. Experimental results provide evidence on the effectiveness of the proposed method which clearly outperforms the conventional approaches currently employed. This method can be extended for designing the Join Cost function, offering many advantages. - The adaption, implementation and the evaluation of a HMM speech synthesis framework for the case of the Greek language. This technology is capable of producing adequately natural speech in terms of intelligibility and intonation, offering many advantages and flexibility in constructing and manipulating general purpose TTS systems. In addition, innovative applications for telecommunication systems and services are described, having TTS technology as a main component.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2010
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Karabetsos, Sotirios
- Καραμπέτσος, Σωτήριος
Subjects
dc:subject × 20- Σύνθεση φωνής
- Ταξινόμηση μιας τάξης
- Επιλογή ακουστικών μονάδων
- Φασματικές αποστάσεις
- Κρυφά μαρκοβιανά μοντέλα
- Κόστος ένωσης
- Τηλεπικοινωνιακές υπηρεσίες
- Κινητά τηλέφωνα
- Speech synthesis
- Unit selection
- Join cost
- One class classification
- Spectral distances
- HMM
- Telecommunication services
- Embedded text to speech
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/24814