Analysis of D/A and A/D Conversions in Quantization-based Audio Watermarking

Abstract

Due to its advantages, quantization-based embedding has been introduced into audio watermarking to improve robustness performance. Existing audio watermarking algorithms often focus on a given attack. However, in some transmission environments, digital audio files may suffer from the different attacks. For example, effects of D/A and A/D conversions (denoted as DA/AD in this paper) on audio watermarking may be modeled as modification of signal energy, phase changes, and noises corruption. These effects present an important challenge to audio watermarking. Extensive experiments show that phase changes of audio signals caused by the DA/AD may be represented as some extent of temporal scaling. Furthermore, we analyze the performance of quantization-based audio watermarking against these attacks in the DA/AD, and present corresponding calculation expressions of BER (Bit Error Rate) in case of Gaussian noise and modification of audio amplitude, and then investigate the influences of temporal scaling caused by the DA/AD on audio watermarking. As a conclusion, audio watermarking quantization-based is very susceptible to the DA/AD.