Many telecommunication applications for text-to-speech synthesis (TTS) Involve speaking names and/or addresses. Two services often considered are a caller-name Identification service (often called "Audible Caller Name Delivery" or "Who's Calling" in the U.S.). and a reverse directory service. This paper will describe our research, development, and recommendations In providing speech synthesis for a heavily used reverse directory service. Issues Involved include highly accurate name pronunciation, high synthesis Intelligibility, and required directory preprocessing.
Keywords: reverse directory service, name pronunciation, speech synthesis