With rapid growth in real-world speech-based transactions via communication networks, the need for a reliable mechanism to prove the credibility of speech content is highly desired. This paper presents a fragile watermarking technique for such purposes. The proposed approach is also useful for speaker verification. It breaks the speech signal into a series of non-overlapping frames and encodes the watermark into those frames consecutively. The watermark sequence for each frame is dependent on the statistical characteristics of the previous frame, therefore any signal discontinuity caused by malicious purposes will be detected and the speaker can be verified as well by the watermark. Experiments showed very encouraging results, including reasonable detection rate even under signal compression and filter attacks.
Cite as: Cheng, Y.-W., Lee, L.-S. (2001) Credibility proof for speech content and speaker verification by fragile watermarking with consecutive frame-based processing. Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001), 1895-1898, doi: 10.21437/Eurospeech.2001-448
@inproceedings{cheng01b_eurospeech,
author={Yiou-Wen Cheng and Lin-Shan Lee},
title={{Credibility proof for speech content and speaker verification by fragile watermarking with consecutive frame-based processing}},
year=2001,
booktitle={Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001)},
pages={1895--1898},
doi={10.21437/Eurospeech.2001-448}
}