Experiments have shown that there is a big difference between noise maskers and tone maskers, and also the characteristics of the masked audio. [5] mostly describes the cases noise-masking-tone and tone-masking-noise. When coding data, the resulting quantization errors are seen as noise, and thus only the tonality of the masker, and not the masked data, has to be estimated. The tonality is used in section 3.2.5 for estimation of the masking threshold.