Intonation: Theory, Models, and Applications
This paper describes a text-to-speech (TTS) application for airline announcements which enables airport staff to specify the type and content of a synthetic announcement. This is a limited domain similar to that for train announcements or other successful restricted-vocabulary systems. Our system demonstrates the possibility of producing natural-sounding prosody automatically for a specified application. Within our chosen domain, we have access to the semantic and pragmatic knowledge which is essential for generating high-quality synthetic prosody. We combine this information with a sophisticated linguistic approach to prosody specification. These two aspects of our system allow it to supersede the unnatural and often unintelligible automatic announcement systems found in many current commercial applications.
Bibliographic reference. Delaney, Martina / Monaghan, Alex I. C. (1997): "SAMBA: prosody in an airline announcement generation system", In INT-1997, 79-82.