Real-time architectures for a network-based text-to-speech service implementation

The communication as we know it is about to change, the simple peer-to-peer voice type of calls being replaced by a mix of different media belonging to the same session. Network-based speech processing technologies are increasingly used, adding flexibility, speed and naturalness to various telecom services. For instance, text-to-speech (TTS) synthesis is already incorporated into commercial telephony applications, but in most of the cases the services offered are not real-time. This paper proposes a solution in building up a network-based media processing environment and designing a multi-threaded and multi-process TTS engine, integrated with the rest of the network by means of flexible and standardized signaling protocols. As a particular application of the proposed approach, we present a convergent real-time telecom service, combining instant messaging, presence and TTS in a single new experience.