Low complexity channel error mitigation for distributed speech recognition over wireless channels

Distributed speech recognition (DSR) has been recently proposed as an efficient way of translating automatic speech recognition technologies to mobile and IP network application. In this paper we propose a channel error mitigation technique with a low computational complexity that improves the mitigation technique proposed in the ETSI standard for DSR (ETSI-ES-201-108 v1.12) for bad channel conditions. We also study the influence of the vector quantization index assignment on the proposed mitigation technique and design a new index assignment that gets some improvement on the proposed technique.