An effect of auditory-filter envelope modulation variability on masking

Abstract

In an earlier report [W. C. Treurniet and D. R. Boucher, J. Acoust. Soc. Am. 106, 2146 (1999)], the threshold for a noise probe masked by a complex masker was shown to be lower when the masker partials were harmonically related than when the frequencies of the partials were perturbed by small amounts. The results suggest that the improved detection may arise from differences in the pattern of modulations across auditory-filter outputs in response to harmonic and inharmonic maskers. That is, envelope modulations are identical across filters for a harmonic masker, but the modulations vary for an inharmonic masker or a noise probe. The variable modulation rates across filters due to the noise probe are easier to detect against the invariant background rates generated by the harmonic masker than against the variable background rates resulting from the inharmonic masker. In support of this hypothesis, the difference in masked threshold was strongly related to the variance of the dominant period of envelope modulations measured across auditory-filter outputs. Further, the masking level differences were minimal when the average variance of modulation amplitude was small, as in the case of a single partial per auditory filter.