An Efficient Tampering Detection and Localization Method for Speech Signals

Abstract

The development in digital technologies and advanced speech analysis/synthesis tools these days enable speech signals to be tampered without leaving any perceptual clues. As an important information carrier, the integrity of speech signals should be strictly confirmed and digital watermarking can be used for this purpose. In this paper, a self-embedding speech watermarking scheme is proposed for detection of unauthorized manipulation (tampering) and localization. Firstly, hash code is generated from original speech signal and then this hash code (as a watermark) is embedded directly into the signal itself without affecting the original quality. Once tampering occurred, the watermark in the tampered segment should be destroyed and thus the receiver can detect the tampered position. Experimental results show that the proposed method provides not only inaudibility and blindness but also fragility against tampering and detects the position of tampering.