Modulation Masking Research Articles

Frequency selectivity in amplitude-modulation detection.

For a broadband noise carrier, the modulation detection threshold for sinusoidal amplitude modulation (the test modulation) is measured in the presence of an additional modulation (the masker modulation). Two traditional approaches for revealing effects of frequency selectivity in the audiofrequency domain are shown to give comparable results in the modulation-frequency domain: (1) a typically peaked modulation-detection threshold pattern when the masker modulation is a fixed narrow band of noise, and (2) an effect of leveling off of the increase of the modulation-detection threshold when, for a fixed test-modulation frequency, the masker-modulation bandwidth is widened beyond a certain ‘‘critical’’ bandwidth. It is argued that the present results on frequency selectivity in modulation detection underline the perceptual relevance of a spectral decomposition of a signal’s temporal envelope and provide a rationale for the application of modern concepts like the speech-envelope spectrum or the modulation-transfer function in relation to speech intelligibility.

Across-critical-band processing of amplitude-modulated tones.

Two experiments using two-tone sinusoidally amplitude-modulated stimuli were conducted to assess cross-channel effects in processing low-frequency amplitude modulation. In experiment I, listeners were asked to discriminate between two sets of two-tone amplitude-modulated complexes. In one set, the modulation phase of the lower frequency carrier tone was different from that of the upper frequency carrier tone. In the other stimulus set, both amplitude-modulated carriers had the same modulator phase. The amount of phase shift required to discriminate between the two stimulus sets was determined as a function of the separation between the two carriers, modulation depth, and modulation frequency. Listeners could discriminate a 50 degrees-60 degrees phase shift between the modulated envelopes for tones separated by more than a critical band. In experiment II, the modulation depth required to detect modulation of a probe carrier was measured in the presence of an amplitude-modulated masker. The threshold for detecting probe modulation was determined as a function of the separation between the masker and probe carriers, the phase difference between the masker and probe modulators, and masker modulation depth (in all conditions, the rate of probe and masker modulation was 10 Hz). The threshold for detecting probe modulation was raised substantially when the masker tone was also modulated. The results are consistent with theories suggesting that amplitude modulation helps form auditory objects from complex sound fields.

Interference of modulation rate discrimination of pure tones

Modulation rate discrimination (Δfm) thresholds were obtained for sinusoidally amplitude modulated (SAM) pure tones. The carrier frequencies were 1000 and 4000 Hz and the modulation rates were 5, 10, 20, and 50 Hz. The Δfm thresholds of one carrier (the probe) were also obtained when the other carrier (the masker) was presented simultaneously. The masker was either SAM or was unmodulated. In all conditions the depth of SAM was 100%. This experiment was designed to determine how much the modulated masker interferes with the discrimination of modulation rate and how this interference changes as a function of the difference in modulation rate between the masker and probe. It has been shown previously [Yost and Sheft, J. Acoust. Soc. Am. Suppl. 1 83, S35 (1988)] that a modulated masker interfered significantly with a listener's ability to detect SAM of the probe when the masker and probe were modulated at the same rate. This modulated detection interference (MDI) was reduced when the masker was modulated at a different rate than the probe. These earlier results suggested that there might be auditory channels “tuned” to different rates of amplitude modulation. The present experiment was designed to further test the concept of AM tuned channels. [This research was supported by the NINCDS.]

Modulation masking patterns

Modulation thresholds for sinusoidally amplitude‐modulated (SAM) broadband noise were obtained for modulation frequencies from 2 to 512 Hz using a two‐interval, forced‐choice adaptive procedure. The noise carrier was on continuously throughout a block of trials, and was modulated for 500 ms in one of the two observation intervals. Thresholds were obtained in quiet and in the presence of a SAM broadband noise masker. In the masking conditions, the same noise carrier, presented at a spectrum level of 15 dB SPL, was used for the signal and the masker. The masker was modulated in both of the 500‐ms observation intervals. The modulation frequency of the masker was 4, 16, or 64 Hz; its modulation depth (m) was 0 (no modulation), 0.5, or 1.0. For a given masker modulation frequency, the modulation masking patterns generally were bandpass, with the greatest amount of masking occurring when the signal and masker modulation frequencies were the same. With a few consistent exceptions, there was a monotonic relation between masker modulation depth and amount of masking (the greater the modulation depth the greater the amount of masking). These data indicate a tuning of the auditory system for the detection of modulation. [Work supported by NIH.]

Binaural modulation detection

A sinusoidally amplitude‐modulated (SAM) noise (modulation depth: 1.0) was used to mask another SAM noise, the signal, in a two‐interval forced‐choice experiment. Signal and masker modulation frequency were both either 4 or 32 Hz, and were added in quadrature. The same wideband noise carrier was used for both masker and signal (spectrum level: 15 dB SPL). The whole stimulus was filtered after modulation (either 4‐kHz low pass or 4‐kHz high pass). Modulation depth (m) of the signal was varied adaptively to track threshold. In the case of diotic presentation (Mo‐So), threshold m was about 0.30 for both modulation frequencies and both filtering conditions. In a binaural condition Mo‐Sm (masker presented to both ears and signal to only one ear) some subjects. showed a release from masking of up to 10 dB (m = 0.10), while others. showed none. The release from masking occurred for both filtering conditions, but only for the lower (4‐Hz) modulation rate. The results will be, discussed in relation to current models of binaural processing and the sluggish nature of the binaural system. [Work supported by NIH.]

Detectability of a pulsed tone in the presence of a masker with time-varying interaural correlation

Detectability of a filtered probe tone (250, 500, or 1000 Hz) was measured in the presence of a narrow-band Gaussian masker centered at the signal frequency. The signal was interaurally phase-reversed (Spi), and the masker's interaural correlation varied sinusoidally between +1.00 (NO) and -1.00 (Npi) at a varaible rate (fm = 0--4 Hz). The signal was presented at various points on the masker's modulation cycle. For 0-Hz modulation (fixed interaural correlation) signal threshold decreased monotonically as the masker's interaural correlation was changed from -1.00 to +1.00 (by a total of about 20, 16, and 8 dB, respectively, for 250-, 500-, and 1000-Hz signals). For fm greater than 0 the function relating signal threshold to the masker's interaural correlation at the moment of signal presentation became progressively flatter with increasing fm for all signal frequencies. For fm = 4 Hz the function was flat; there was no measurable effect of masker interaural correlation on signal detectability. Estimates of minimum binaural integration time based on these data ranged from 44--243 ms, supporting previous studies which have noted the binaural system's relative insensitivity to dynamic stimulation. Additionally, the estimated time constants were approximately twice as large at 250 Hz as at 500 Hz, indicating observers could follow binaural fluctuations better at 500 Hz. The time-constant estimates at 1000 Hz were not suggiciently reliable to permit comparisons with the lower-frequency data.