A Study On Detection And Recovery Of Speech Signal Tampering

In this study, we use fragile self-recovery audio watermarking technology, aiming at audio signal protection. The so-called protection means to embed information in original signals, when the audio signal is tampered, the watermarked signal can be recovered by the information embedded in advance. The same algorithm is rare in the audio field. Most research focuses on the use of different types of instruments, such as higher-order spectrum analysis to detect the damaged portion of the audio signals or to use digital watermarking to protect ownership of the audio signal. As for the recovery of the data, only a few methods is studied and these methods are prime, are not able to protect against the practical complex attack scenarios.In this paper, a novel imperceptible, fragile and blind watermark scheme is proposed for speech tempering detection and self-recovery. We embed an approximate version of the original signal into the least significant bits(LSBs). One can accurately localize the tampered area and recover the original signal with the watermark. The watermark information is shared n a frames-group instead of stored in one frame. The scheme trades off between the data waste problem and the tampering coincidence problem. Based on the embedded watermark, one can estimates the original signal by solving linear equations by the least square method. The result of a subjective evaluation of listening test shows that the intelligibility of recovered speech is 83.7% when 20% part has been totally tampered.