Audio watermarking for Internet

Nowadays the MP3 (MPEG 1 layer 3) normally used in the Internet crops the inaudible signal components that watermarks are often embedded. This paper proposes a method to embed a watermark in the audio signal, which is robust against the MP3 compression, and the sound is maintained according to the human audio ability. Since MP3 is a lossy compression technique that crops the inaudible frequency ranges of signal, we can not embed the watermark in the inaudible signal ranges. In this method, we select the audible frequency ranges with small sound pressure level as watermark embedding frequency range. By this technique the embedded watermark is robust against MP3, and the human will not recognize the watermark data as noise. In addition, we try to utilize the color image as the watermark that is convenient to claim the ownership. We performed the experiments with the popular songs and the results are satisfied.