Identifying whether a phone call comes from VoIP (Voice over Internet Protocol) is a challenging but less-investigated audio forensic issue. As shown in a previous study, existing feature based methods do not work well. In this paper, we propose a robust data-driven approach, called CNN-MLS (convolutional neural network based multi-domain learning scheme), to distinguish VoIP calls from mobile phone calls. To better explore the differences between VoIP and mobile phone calls, we first process data with high-pass filtering, and then extract deep features from both temporal domain and spectral domain. Two CNN architectures are designed for accepting data from respective domains, and some tricks such as auxiliary classifiers and individual subnet training are used for accelerating network convergence. The deep features are finally fused in a classification module for identifying the phone call type. The proposed method is evaluated on VPCID (VoIP Phone Call Identification Database) dataset, under various testing conditions. We pay particular attention to tests on data belonging to a source mismatched with the training sources. Experimental results show that, compared with existing methods, our method can achieve satisfactory and better accuracy on two-second-long inputs, implying that an alert may be activated shortly after a VoIP call is made.

Watermarking is a solution for copyright protection and forensics tracking, but recapturing and de-synchronization attacks may be used to effectively remove audio watermarks. Although much effort has been made in recent years, the robustness of audio watermarking against recapturing and de-synchronization attacks is still a challenging issue. Specifically, we first construct the frequency-domain coefficients logarithmic mean (FDLM) feature of digital audio. By theoretical analysis, we conclude that the residual of the two groups' FDLM feature is robust against recapturing attack. We then propose a robust audio watermarking method based on this feature using the patchwork framework. Compared with the method having the best robustness performance against recapturing attack, the BER value of our method is decreased by 7%. Besides that, the proposed method outperforms the state-of-the-art patchwork-based watermarking methods notably, under recapturing and post-processed with signal processing operations and de-synchronization attacks.