This paper presents a brief overview of a new duration model for General American English, which I am currently developing in connection with work being done at both Cornell University and Eloquent Technology, Inc. in speech synthesis by rule. In addition to pursuing the standard goals of synthesis rule development, such as high intelligibility and naturalness, we strive in this work to develop a linguistically realistic system with which we can gain insights into the timing structure of English and other languages. The paper expands upon Hertz (1991), which describes and motivates a model of speech timing called the phone-anti-transition model. While the phone-and-transition model provides an appropriate framework for synthesis of any language or dialect (see Hertz, 1990a), this paper will focus on its application to General American English (henceforth "GA"). presenting the basic rule algorithm for determining the internal timing structure of stressed syllable nuclei in GA. The first section gives a brief outline of the phone-and-transition model. The second section describes the methodology being used to develop the duration model for GA. The third section discusses the duration model, focussing on the rules that compute the durations of the sub-units of stressed syllable nuclei. A final section provides some concluding remarks.

This paper is copyrighted, and is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) - see https://creativecommons.org/licenses/by-nc-nd/4.0/