Advances in automatic speech recognition (ASR) technology promise new products and services that capitalize on the enhanced capability of human‐machine communication. This paper: Addresses several key issues in the design of a deployable ASR, Discusses new dimensions and developments of the technology, Presents several examples of successful applications of the ASR technology in AT&T's products and services, and Describes the research effort needed to realize many new telecommunications applications in years to come.

The voice processing equipment market has expanded to include both large and small businesses, and the influence of voice processing—evidenced in voice mail messages and voice response systems—has become widespread. AT&T systems such as the Conversant® Voice Information System and the AUDIX voice messaging system have been designed to enhance both business and personal telecommunications. This paper explores the architectures of AT&T's Conversant and AUDIX systems in terms of hardware, software, and application design.

No abstract

No abstract

A flexible and cost‐effective system for introducing speech technology into the marketplace is described. The system answers telephone calls from lines and trunks and provides verbal prompts and information to the caller. The caller can request specific information and direct the transaction by using either touch‐tone or voice input. Requested information is obtained either from a small data base on the system disk or from an external source. Other system functions include recognition of isolated words and connected digit strings, recording and encoding of phrases and announcements for playback during a transaction, and call transfer and origination. A special interpreted language for specifying the transaction dialog allows diverse applications of the system to be easily written and modified.