In this paper we have presented a robust watermarking procedure for digital audio. The procedure embeds a watermark imperceptibly in the time domain exploiting the temporal and simultaneous masking properties of the human auditory system (HAS). The spectral characterization of the audio is done using the short time Fourier transform (STFT), which provides a better description of non-stationary signals when compared to using the discrete Fourier transform. The watermark is a perceptually shaped pseudorandom sequence, which is used to modify the audio samples. The original audio is not required during the extraction of the watermark, except when the audio signal is subjected to cropping attack. The watermarking procedure exhibits resilience towards several audio distortions occurring from MPEG 1 Layer 3 coding, colored noise, cropping, temporal resampling and jittering. © 2005 JD.