Methods for expressing spectral modulation depth and the effects of modulator shape on spectral modulation detection thresholds

The detection of sinusoidal modulation is commonly used for assessing the auditory perception of temporal, spectral, and spectro-temporal acoustic features. For temporal (amplitude) modulation, the sinusoidal modulator usually is expressed on a linear amplitude scale. For spectral modulation, the modulator has been specified to be sinusoidal on a linear amplitude scale (lin-sin), analogous to temporal modulation, or to be sinusoidal on a logarithmic amplitude scale (log-sin), with the notion of approximating a sinusoidal excitation pattern. The definition of modulation depth depends on the measurement scale, the chosen reference points (i.e., midpoint-to-peak, peak-to-valley), and the order of operations when expressing depth in dB. Such differences can make it difficult to compare spectral modulation detection results among investigations. Several analytic methods for equating modulation depth across the different scales and reference points were explored and a preferred method was identified. A complete matrix of values for translating among different modulator shapes and modulation depth definitions is provided. To investigate potential perceptual differences between log-sin and lin-sin shapes, spectral modulation detection was measured for 9 normal-hearing listeners for a range of modulation frequencies. The resulting spectral modulation transfer functions revealed statistically significant effects of both modulator shape and spectral modulation frequency.