Fragile speech watermarking for content integrity verification

Two fragile speech watermarking schemes for content integrity, i.e. exponential-scale odd/even modulation and linear additive watermarking, are proposed and compared with respect to their ability in detecting localized malicious alterations and tolerating content preserving operations. Both algorithms embed watermarks in the DFT magnitude domain by using a simplified masking model to achieve inaudibility. Statistical analysis is performed to identify the detection failure rate and the false detection rate of these two algorithms. These two schemes require no auxiliary authentication data to perform integrity verification, and could distinguish various content preserving operations from malicious tampering. The exponential-scale odd/even modulation is better in detecting localized content alteration while linear additive watermarking is more tolerant to low bit-rate speech coders such as the CELP coders.