IDEAW: Robust Neural Audio Watermarking with Invertible Dual-Embedding

TL;DR

A dual-embedding wa- termarking model for efficient locating and the impact of the attack layer on the invertible neural network in robustness training is considered to enhance both its reasonableness and stability.

Abstract

The audio watermarking technique embeds messages into audio and accurately extracts messages from the watermarked audio. Traditional methods develop algorithms based on expert experience to embed watermarks into the time-domain or transform-domain of signals. With the development of deep neural networks, deep learning-based neural audio watermarking has emerged. Compared to traditional algorithms, neural audio watermarking achieves better robustness by considering various attacks during training. However, current neural watermarking methods suffer from low capacity and unsatisfactory imperceptibility. Additionally, the issue of watermark locating, which is extremely important and even more pronounced in neural audio watermarking, has not been adequately studied. In this paper, we design a dual-embedding watermarking model for efficient locating. We also consider the impact of the attack layer on the invertible neural network in robustness training, improving the model to enhance both its reasonableness and stability. Experiments show that the proposed model, IDEAW, can withstand various attacks with higher capacity and more efficient locating ability compared to existing methods.

Figures (6)

• Figure 1: (a) Pipeline of robust neural audio watermarking. (b) Embedding strategy of existing methods. (c) Dual-embedding strategy of IDEAW.
• Figure 2: Architecture of IDEAW and the training objectives.
• Figure 3: Structure and forward/backward processes of the invertible block.
• Figure 4: Waveforms of (a) host audio (foreground) and watermarked audio (background), (b) host audio and tenfold-magnified residual caused by watermarking, (c) local details (100 points) of the (a) and (b). The left audio is low-energy speech audio while the right is high-energy music audio.
• Figure 5: Linear-frequency power spectrograms of low-energy speech audio (left) and high-energy music audio (right). (a) the host audio, (b) the watermarked audio.