FACE is an ongoing project in CTIF, involving SMC, WING, CSys and A&P divisions. With the deployment of always-on communication terminals and the growing availability of information retrieval services, the challenge of this project is to provide users with adaptive and flexible access to a given service via a variety of networks via user-friendly interfaces. Speech is obviously the most natural interface for communication. Within this project Speech and Multimedia Communications is targeted towards speech processing in the presence of networks and the integration of conversational interfaces in network environments. The emphasis is on the establishment of speech technology modules distributed in networks and adapted to the qualities of involved networks, aimed at providing end-users an overall optimising of the 'Perceived' QoS. The research has focused on robust distributed speech recognition (DSR) in the presence of channel transmission errors in IP- and wireless networks. Two algorithms have been proposed. The first is a method called subvector based error concealment (EC). Conventional EC algorithms for DSR share a common characteristic namely the fact of conducting EC at the vector (or frame) level. This strategy, however, fails to effectively exploit the error-free fraction left within erroneous vectors where a substantial number of subvectors often are error-free. A novel EC approach is proposed for DSR encoded by split vector quantization (SVQ) where the detected erroneous vectors are submitted to a further analysis at the subvector level. Specifically, a data consistency test is applied to each erroneous vector to identify inconsistent subvectors. Only inconsistent subvectors are replaced by their nearest neighbouring consistent subvectors whereas consistent subvectors are kept untouched. Experimental results demonstrate that the proposed algorithm in terms of recognition accuracy is superior to conventional EC methods having almost the same complexity and resource requirement. The second is a method called partial splicing. A technique for mitigating the effect of packet loss in the context of DSR is presented. The proposed packet loss concealment (PLC) technique substitutes packet loss partly by a repetition of neighbouring packets and partly by a splicing in which a number of packets are dropped. Experimental results demonstrate that the proposed PLC technique outperforms existing techniques.
|Effective start/end date||19/05/2010 → 31/12/2017|