ABSTRACT

After years of productive research, speech synthesis is now profitably automating services by answering queries via constrained dialogs, directly accessing individual computer databases, and speaking text created from disparate sources of information. The most successful of these services are carefully designed, in terms of proper selection of synthesizer, and answering a true need for information, but, especially, adequately handling problematic situations inherent to the service. This chapter illustrates some of the components of success for the latter case, inherent service problems: methods to successfully cope with problems with real-world text, synthesizers, and users. The topics covered are the problem areas of inadequate preparation of database text; low pronunciation accuracy of text (names, abbreviations and acronyms); communication of visual formatting aspects; and poor synthesis, especially of intonation that leads to low comprehension. Successful implementations of text-to speech synthesis, in spite of user opinions of unnatural quality for synthesized speech, confirm the value of efforts that overcome these hurdles.