New European standards for speech recognition
The performance of speech recognition systems tends to deteriorate over telephone lines, when other people are speaking at the same time somewhere near, or when the speaker has an accent. The SMADA project has actively contributed to the definition of a new European standard for a noise-robust front-end for Distributed Speech Recognition (DSR). The standard substantially improves the robustness and accuracy of Automatic Speech Recognition (ASR) under real world conditions, and enables distributed speech recognition over mobile networks. DSR can, by using an efficient noise-reduction method, improve the quality of the speech signal. It lowers the influence of noisy environments, bandwidth limitations, coder/decoder (codec) effects and transmission errors, which all tend to decrease the performance of the recognition system. The formal standard - a major achievement of the project - was published by the European Telecommunications Standards Institute (ETSI) in October 2002. The standard is currently being discussed and is under evaluation in the Third Generation Partnership Project (3GPP) for implementation as a default codec component for speech-enabled services in Third Generation (3G) wireless networks. This outcome of the SMADA project could prove of value in the near and mid-term future when speech-enabled and multimodal services become available in 2.5G and 3G mobile networks.