An efficient content authentication scheme in encrypted speech based on integer wavelet transform

Qing Qian, Hongxia Wang, Canghong Shi, Huan Wang

Abstract: With the advent of cloud storage, the cloud service brings a lot of convenience, but it also brings some secure issues. In order to realize the confidentiality and integrity of speech content stored in the cloud, an efficient watermarking based on integer wavelet transform in encrypted domain is proposed. Firstly, the original speech is scrambled by chaotic mapping. Then, the integer wavelet approximate components of each scrambled speech frame are encrypted by block cipher. Subsequently, the encrypted components are used to generate watermark which is embedded into the detail components. The process of watermark extraction and authentication is responsible for detecting attacks and locating tampered area. The extracted watermark is used for self-synchronization and tampering detection simultaneously. Theoretical analysis and simulation results show that the proposed watermarking has high security, better inaudibility and strong ability to locate the tampered area.