Quantization Index Modulation audio watermarking system using a psychoacoustic model

We propose a system in which the watermark message is embedded into audio signal by means of Quantization Index Modulation (QIM) scheme. However, the method is sensitive to gain scaling attack. We use a value-metric scaling invariant method called Rational Dither Modulation (RDM) to cope with this problem. At the same time, we exploit the psychoacoustic model to improve the audio signal's perceptual quality after the watermark embedding procedure. In comparison of the traditional QIM which use a fixed quantization step, our method adjusts the quantization step with psychoacoustic model adaptively. Experimental results show that our watermarking scheme is robust against common signal processing attacks while keeping the watermarked audio a high perceptual quality.