Speech content authentication integrated with CELP speech decoders

Abstract

A speech content authentication scheme, which is integrated with CELP speech coders to minimize the total computational cost is proposed in this research. Speech features relevant to semantic meaning are extracted, encrypted and attached as the header information. A low cost synchronization algorithm is used to resolve mis-synchronization caused by content preserving operations. Silent and tonal regions in the speech are identified with algorithms of low complexity to enhance the precision of integrity verification. This scheme is not only much faster than traditional cryptographic bitstream integrity algorithms, but also more compatible for a variety of applications. Experimental results are collected by using the GSM-AMR speech coder, and statistical analysis is performed to calculate the false positive rate of tamper detection.