The future role of text to speech synthesis in automated services

This paper suggests that the next generation of synthesis systems will inevitably take more account of the type of information being presented to them; that the interfaces to such systems will become more generic, and that the type of processing conducted as part of the synthesis process will become more diffuse and data orientated. The paper also suggested that advances in speech synthesis can best be achieved when developed within a complete multi-modal spoken language framework. BT is developing a robust multi-modal spoken language framework within the Maya project. Maya is composed from a set of co-operating services. These services come in two forms, core services (facilitators) and component services. The facilitating services help to control the flow of information about the system and co-ordinate the behaviour of component services. The number and type of component services available within a particular configuration of Maya will vary depending on the modality of the application being built. As a minimum, the Maya system will contain speech recognition, parsing, dialogue and speech generation services. Each of these components services may in turn be composed of a number of co-operating sub-processes. As an example, the speech generation component within the Maya has at its core, BT's Laureate text to speech system.