Dissertation/Thesis Abstract

Speech Synthesis Using Unsupervised Learning
by Datta, Aditi, M.S., California State University, Long Beach, 2017, 215; 10639160
Abstract (Summary)

This thesis introduces a general method for incorporating the distributional analysis of textual and linguistic objects into text-to-speech (TTS) conversion systems. Conventional TTS conversion uses intermediate layers of representation to bridge the gap between text and speech. Collecting the annotated data needed to produce these intermediate layers is a far from a trivial task, possibly prohibitively so for languages in which no such resources are in existence. Distributional analysis, in contrast, proceeds in an unsupervised manner and so enables the creation of systems using textual data that are not annotated. The method, therefore, aids the building of systems for languages in which conventional linguistic resources are scarce but is not restricted to these languages.

The distributional analysis proposed here places the textual objects analyzed in a continuous-valued space, rather than specifying a hard categorization of those objects. This space is then partitioned during the training of acoustic models for synthesis, so that the models generalize over objects’ surface forms in a way that is acoustically relevant.

The method is applied to three levels of textual analysis: to the characterization of sub-syllabic units, word units, and utterances. The entire system was built with no reliance on manually labeled data or language-specific expertise. Results of a subjective evaluation are presented.

Indexing (document details)
Advisor: Tsang, Chit-Sang
Commitee: Yang, Hengzhao, Yeh, Hen-Geul (Henry)
School: California State University, Long Beach
Department: Electrical Engineering
School Location: United States -- California
Source: MAI 57/04M(E), Masters Abstracts International
Source Type: DISSERTATION
Subjects: Language, Artificial intelligence, Computer science
Keywords:
Publication Number: 10639160
ISBN: 978-0-355-61524-1
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest