In this paper, we describe the design procedure for a wireless communication interactive voice response system. The application must work in a very noisy environment which has imposed many design constraints. We will address the sensible aspects of three components of the application: the voice activity detector (VAD), the automatic speech recognition (ASR) system, and the confidence measure (CM) determination. In order to get a satisfactory product, it has been necessary to reduce the important mismatch between available linguistic and acoustic resources and the operational environment. Adaptation techniques for the acoustic models of the speech recognition system have proven to be effective to speed up the application deployment time.
展开▼