The invention relates to a method for embedding a watermark in an audio signal, according to which a spectral representation of the audio signal and a spectral representation of the watermark signal are first determined (14, 16). The spectral representation of the watermark signal is then processed, based on a psychoacoustic masking threshold (24) of the audio signal (22). The processed watermark signal is combined with the audio signal (18), to obtain an audio signal furnished with a watermark. The spectral representation of the watermark signal is processed iteratively in the following way: a predetermined watermark initial value (26) is first selected; the interference that has been introduced into the spectral representation by the predetermined watermark initial value after a quantization of the spectral representation of the audio signal, is determined (28). If the interference introduced by the watermark initial value is greater than the predetermined interference threshold (32), the watermark initial value is then modified until said interference introduced into the spectral representation of the audio signal by a modified watermark initial value after quantization, is less than or equal to the predetermined interference threshold. The modified watermark initial value that is obtained at the end of the iteration is used as a processed watermark signal to be combined with the audio signal. It is therefore no longer possible to weaken or remove a watermark by quantization. The invention provides complete control over the energy of the watermark, in order to ensure that a watermark is embedded into an audio signal with either the best possible degree of detectability or the best possible audio quality.