Prosodic Annotation in a Thai Text-to-speech System

Prosodic Annotation in a Thai Text-to-speech System* Siripong Potisuk Department of Electrical and Computer Engineering The Citadel, The Military College of South Carolina 171 Moultrie Street Charleston, South Carolina 29409 USA siripong.potisuk@citadel.edu Abstract. This paper describes a preliminary work on prosody modeling aspect of a text-to- speech system for Thai. Specifically, the model is designed to predict symbolic markers from text (i.e., prosodic phrase boundaries, accent, and intonation boundaries), and then using these markers to generate pitch, intensity, and durational patterns for the synthesis module of the system. In this paper, a novel method for annotating the prosodic structure of Thai sentences based on dependency representation of syntax is presented. The goal of the annotation process is to predict from text the rhythm of the input sentence when spoken according to its intended meaning. The encoding of the prosodic structure is established by minimizing speech disrhythmy while maintaining the congruency with syntax. That is, each word in the sentence is assigned a prosodic feature called strength dynamic which is based on the dependency representation of syntax. The strength dynamics assigned are then used to obtain rhythmic groupings in terms of a phonological unit called foot. Finally, the foot structure is used to predict the durational pattern of the input sentence. The aforementioned process has been tested on a set of ambiguous sentences, which represents various structural ambiguities involving five types of compounds in Thai. Keywords: Text-to-speech, Prosody. 1. Introduction Presently, widespread use of text-to-speech technology is limited by its inability to produce high- quality speech. That is, intelligibility and naturalness of synthetic speech is still not quite at the level acceptable by human listeners. In particular, the naturalness issue can be attributed to the lack of sophisticated prosody-generating scheme. Prosody is often described as a suprasegmental feature of speech (a term for describing phonological features of those aspects of speech that involve more than single consonants or vowels). Acoustically speaking, prosody can be defined as change in the fundamental frequency (F0), timing, and amplitude of a speech signal. Speakers control the prosody of an utterance in order to signal linguistic and affective information. Linguistic prosody is used by speakers to signal * The author would like to thank the Citadel Foundation for its financial support in the form of a presentation grant. Copyright 2007 by Siripong Potisuk 405 grammatical information at the syllable, word, or sentence level (e.g., stress, intonation). Affective prosody, on the other hand, is used to convey information that indicates speaker’s intentions, attitudes, or emotional states. In addition to linguistic and affective information, prosody can also be used to convey non-linguistic information concerning speaker’s personal characteristics such as age, gender, idiosyncrasy, speaking style, and physical condition. Such characteristics may or may not be under the speaker’s volitional control. It is part of the intelligibility and naturalness of his/her speech. This paper will deal with linguistic prosody only. The role of linguistic prosody in spoken language is similar to that of punctuation in written language. Punctuation is used …

Content truncated to 500 words