Speech Technology in Reality — Applications, Their Challenges and Solutions

Abstract

Speech technology is one of the very first technologies, which seems to offer real user friendly access to human like systems. A wide field of applications should become possible. But reality is still quite different. Automatic speech recognition is still far from human capabilities and even speech synthesis does not yet offer human like quality. So, reality of applications is still in a rather basic status. Background noise, room echoes, changing frequency characteristics of electric and acoustic transmission, speaker and articulation variations are only few aspects of complexity in speech recognition. Only through new and advantageous methods for cancellation of background noise and room echoes, equalization of transmission channels and adaptation to varying speakers and articulation new applications became attractive. Such interesting applications are in the car, in consumer applications, in advanced telephone voice interaction and in many other areas, where speech is an interesting medium for user friendly man-machine interaction. The dialog aspect will play a growing role and therefore especially in noisy environment further enhancements of speech output will play an important role too.We have to care for high intelligibility in such environment to make really good and efficient applications.