Audio Watermarking Using Beat Detection and Pitch Estimation

This paper presents a novel semi-blind Audio Watermarking Algorithm in the Time Domain with the use of Beat detection and Pitch Estimation. Using the estimated values of the beat locations and their respective pitch or fundamental frequencies, a watermark is generated from the binary information that is to be hidden, which is then embedded in the audio file. The robustness of this algorithm has been gauged using standard audio processing attacks such as Resampling, Re-quantisation and Noise attack using AWGN. Apart from these, the watermarked audio file has been subjected to equalisation attacks which are prevalent in the music industry today, using equalisation modes such as Jazz, Rock, Classical and Bass. The performance of the algorithm has been proved to be satisfactory against all such attacks.