This paper evaluates degradation of the audio content for the perceptible removable watermark. Watermarking is done in the spectral domain. The frequencies for watermark embedding are chosen according to a pseudo-random sequence making the method robust. Consequentially, the lower quality audio can be used for promotional purposes. For a fee, the watermark can be removed with a secret watermarking key. Objective and subjective testing was conducted in order to measure degradation level for the watermarked music samples and to investigate residual distortion for different parameters of the watermarking algorithm and different genres.