Audio watermarking

Abstract

The copying of digital files has become a very easy task. Almost all media can be, and is represented digitally and distributed across the world in a matter of a few mouse clicks. A novel audio watermarking scheme is presented to inaudibly encode an owner's signature into an audio file in WAY format. This non-blind watermarking algorithm is based on both time and frequency base modulation. Segments of the file are compressed or expanded by adding or removing samples in the frequency domain. The amplitude of these samples is below the threshold limits calculated with Psychoacoustic Model I. This is manifested in the time domain as the lengthening and shortening of segments of the audio signal. Information is represented by this pattern of expansion and compression. The proposed method achieves a balance between the aural imperceptibility of the introduced distortion and the robustness of the watermark. Results show successful recovery of the signature which allows conclusive authentication of the owner even after attacks like filtering, cropping, MPEG compression, resampling and the addition of noise.