REPAIR: fragile watermarking for encrypted speech authentication with recovery ability

Abstract

Confidentiality and integrity are fundamental requirements when transmitting and storing data. In order to guarantee the confidentiality and integrity of speech signal, we present a novel wateRmarking scheme for Encrypted sPeech AuthenticatIon with Recovery (REPAIR) scheme. In REPAIR, an encryption algorithm is first designed based on a hyper-chaotic method to improve the confidentiality of the original speech by encryption. Subsequently, a watermark generation and embedding algorithm is proposed to generate and embed the check bits and compression bits. Afterwards, a content authentication and tampering recovery algorithm is introduced to locate and recover the tampered speech frames. Meanwhile, a speech decryption algorithm is also presented to decrypt the encrypted speech. Analysis and experimental results demonstrate that REPAIR can detect and locate synchronization attacks and de-synchronization attacks without using the auxiliary synchronous code. Additionally, REPAIR can also recover the tampered content with high quality.