On Customizing Prosody in Speech Synthesis: Names and Addresses as a Case in Point

ON CUSTOMIZING PROSODY IN SPEECH SYNTHESIS: NAMES AND ADDRESSES AS A CASE IN POINT Kim E. A. Silverman Artificial Intelligence Laboratory NYNEX Science and Technology, Inc. 500 Westchester Avenue White Plains, New York 10604 1. ABSTRACT This work assesses the contribution of domain-specific prosodic modelling to synthetic speech quality in a name-and-address information service. A prosodic processor analyzes the textual structure of labelled input strings, and inserts markers which spec- ify the intended prosody for the DECtalk text-to-speech synthe- sizer. These markers impose discourse-level prosodic organization, annotate the information structure, and adapt the speaking rate to listeners in real time. In a quantitative comparison of this domain-specific modelling with the default rules in DEC- talk, the domain-specific prosody was found to reduce the tran- scription error rate from 14.6% to 6.4%, reduce the number of repeats requested by listeners from 2.6 to 1.1, and to sound signif- icantly easier to understand and more natural. This result demon- strates the importance of prosodic modelling in synthesis, and implies an even more important role for prosody in more compli- cated domains and discourse structures. 2. INTRODUCTION Text-to-speech synthesis could profitably be used to auto- mate or create many information services, if only it were of better quality. Unfortunately it remains too unnatural and machine-like for all but the simplest and shortest texts. It has been described as sounding monotonous, boring, mechanical, harsh, disdainful, peremptory, fuzzy, muffled, choppy, and unclear. Synthesized isolated words are rela- tively easy to recognize, but when these are strung together into longer passages of connected speech (phrases or sen- tences) then it is much more difficult to follow the meaning: the task is unpleasant and the effort is fatiguing [1]. This less-than-ideal quality seems paradoxical, because published evaluations of synthetic speech yield intelligibil- ity scores that are very close to natural speech. For exam- ple, Greene, Logan and Pisoni [2] found the best synthetic speech could be transcribed with 96% accuracy; the several studies that have used human speech tokens typically report intelligibility scores of 96% to 99% for natural speech. (For a review see [1]). However, segmental intelligibility does not always predict comprehension. A series of experiments [3] compared two high-end commercially-available text-to-speech systems on application-like material such as news items, medical benefits information, and names and addresses. The result was that the one with the significantly higher segmental intelligibility had the lower comprehension scores. Although there may be several possible reasons for seg- mental intelligibility failing to predict comprehension, the current work focuses on the single most likely cause: syn- thesis of prosody. Prosody is the organization imposed onto a string of words when they are uttered as connected speech. It includes pitch, duration, pauses, tempo, rhythm, and every known aspect of articulation. When the prosody is incorrect then at best the speech will be difficult or impossible to understand [4], at worst listeners will be mis- understand it with being aware that they have done so. Arguments for the importance of prosody in language abound …

Content truncated to 500 words