Masker Duration Research Articles

Comodulation masking release: The effect of the characteristics of noise bands presented before and after a signal

CMR was measured as a function of masker duration in three conditions. In the first condition, a 400-ms pure-tone signal was presented in the temporal center of comodulated masking bands ranging in duration from 400 to 1200 ms. No effect of masker duration was observed: CMR was equivalent to that found in continuous maskers. In the second condition, continuous independent bands identical in spectra to the comodulated bands were gated off before, and on following, the comodulated bands. The presence of the independent bands reduced CMR considerably. Increasing the duration of the comodulated bands increased CMR somewhat; however, CMR was still substantially smaller than that observed without the independent bands. In the third condition, band limited noise spanning the same range as the comodulated bands was substituted for the independent bands. The bandlimited noise had little effect on CMR. In a second experiment, independent bands were gated on for various durations preceding spectrally identical 400-ms comodulated bands masking a 400-ms signal. As the duration of the independent bands increased, CMR decreased. Based on these results, it appears that CMR is not entirely the result of processing only information concurrent with the signal. [Work supported by NIH-NIDCD and Louisiana Board of Regents.]

Effect of temporal gaps on informational and energy-based masking

The effect of a temporal gap on detecting a 1000-Hz tone in the presence of 60-dB SPL simultaneous maskers was examined. Ten-component, random-frequency maskers and broadband-noise maskers were used in a 2-AFC adaptive task. Random-frequency components were drawn from 300 to 3000 Hz, excluding a 160-Hz band around the signal. Temporal gaps of 10, 20, 40, 80, and 160 ms were tested, positioned either at the onset, center, or offset of either the signal or the masker. Without gaps, both signal and masker durations were 200 ms. To maintain equal energy across all conditions, level compensation was applied when gaps were employed. For temporal gaps in either the multicomponent or noise masker, masked thresholds consistently decreased as gap duration increased. Gaps in the masker appeared to provide a temporal window for detection of the signal. However, for gaps in the signal, masked thresholds decreased with the multicomponent masker, but remained constant with broadband noise masker. With multicomponent maskers, the gaps appeared to reduce informational masking by perceptually segregating the signal from the masker. With broadband noise maskers, there was little informational masking and therefore the temporal gaps did not improve performance. [Work supported by NIDCD.]

Psychophysical suppression as a function of signal frequency

Psychophysical suppression was investigated across signal frequency (250, 500, 1000, 2000, and 4000 Hz) in a forward-masking paradigm. Masker duration was 200 ms, signal duration was 20 or 40 ms, and signal delay was 0 or 20 ms. When using a noise masker (spectrum level of 40 dB), the amount of suppression was determined by subtracting threshold in the presence of a broadband masker from that in the presence of a critical band masker. When using a tonal masker (masker level of 50 dB SPL, suppressor level of 70 dB SPL, with the suppressor frequency being 1.2 times the masker/signal frequency), the amount of suppression was determined by subtracting the threshold in the presence of the masker plus suppressor from that in the presence of masker alone. For both masker types, the amount of suppression increased as signal frequency increased up to 1000 Hz, but then reached an asymptote or decreased somewhat as signal frequency increased to 4000 Hz. [Work supported by NIDCD.]

Combinations of forward and simultaneous masking, revisited

In previous work, it was found that combinations of a long duration, broadband noise simultaneous masker and a 1000-Hz sinusoidal forward masker can produce as much as 14 dB more masking of a 1000-Hz signal than would be predicted on the basis of a linear power summation of the individual effects of the two maskers. In this study, the effects of masker and signal bandwidth and of simultaneous masker duration on this excess masking were explored. Using a 20-ms sinusoidal signal and a combination of 400-ms broadband-noise simultaneous masker and a 300-ms sinusoidal forward masker, the original effect in a new group of four subjects was replicated. When independent broadband noises served as both maskers and as the signal, and when the simultaneous masker was the same duration as the 20-ms signal, excess masking was completely eliminated. Longer duration simultaneous maskers, a sinusoidal signal, or a sinusoidal forward masker all produced excess masking to varying degrees. No existing model of the additivity of masking provides a good account of these effects. [Work supported by NIH.]

Psychophysical evidence of excitatory connections across auditory frequency channels.

A simultaneously masked auditory signal becomes easier to detect as its onset is delayed from masker onset. This ‘‘enhancement effect’’ depends upon masking components presented outside of the signal’s frequency channel, indicating the participation of an across‐channel process. The nature of this process has not been clearly established due to the difficulty of separating the influence of excitation and inhibition in simultaneous masking. Here, the forward‐masked threshold of a 20‐ms, 1000‐Hz tone was determined in the presence of one or two masking tones [M1 (1000 Hz, 50 dB) and M2 (variable from 500 to 3000 Hz, 70 dB)]. When the masker duration was 500 ms, adding M2 to M1 reduced signal threshold when M2 was presented at 1150 Hz (suppression), but had no influence on performance when M2 was above about 1300 Hz. In contrast, when the masker duration was 40 ms, adding M2 to M1 did not lower threshold at 1150 Hz, and always increased threshold when M2 was presented above 1300 Hz. This latter threshold inc...

Temporal masking of amplitude modulation detection with a wideband noise carrier

Thresholds were measured for detecting amplitude modulation (AM) of a gated wideband noise signal that was either immediately preceded or followed by a modulated noise masker. The signal AM rate was 2.5, 5, 10, or 20 Hz and the masker AM rate ranged from 2.5 to 40 Hz. Signal duration was 200 or 400 ms and masker duration was 400 ms. The maximum elevation in signal AM detection thresholds due to the presence of masker modulation ranged across subjects from roughly 4 to 10 dB with the amount of masking generally greatest when the signal and masker AM rates were either close or the same. The extent of tuning in the modulation domain was broad in both the forward and backward masking conditions. When not modulated, the masker can be viewed as a fringe of the signal carrier. Separate conditions evaluated the effects of carrier fringes on signal AM detection. At low AM rates, the addition of either the forward or backward fringe benefitted AM detection. These results suggest that the decrement in low-rate AM detection due to carrier gating is not solely attributable to the neural onset response and short-term adaptation. [Work supported by NIH and NSF.]

Forward‐masked intensity discrimination: Duration effects

Forward‐masked intensity discrimination shows elevated just‐noticeable differences (jnd’s), for mid‐level standards, a finding argued to be consistent with sensory adaptation [F. Zeng and C. W. Turner, 782–787 (1992)]. Accordingly, the size of jnd’s should increase with increases in masker duration. We tested this notion by measuring forward‐masked jnd’s as a function of level for 1000‐Hz tones in four listeners with normal hearing. Jnd’s were measured for 10‐ms standards in quiet and in the presence of a 1000‐Hz masker with either a 10‐ms or a 250‐ms duration. The masker level was fixed at 90 dB SPL and the interval between masker offset and standard onset was 100 ms. Forced‐choice procedures were employed. The results show significantly less masking for the longer duration masker (250 ms) than for the short one (10 ms). These results are contrary to the results reported for forward‐masked detection. Listening bands revealed by measuring jnd’s as a function of masker frequency suggest that off‐frequency listening may be responsible for this finding [Work supported by the Bryng Bryngelson Fund.]

Modeling the additivity of nonsimultaneous masking

Thresholds were measured for detecting a brief 6-kHz sinusoidal signal preceded by a broadband noise masker (forward masking), followed by the masker (backward masking), or both preceded by and followed by the masker (combined masking). The masker-signal interval was systematically varied. Consistent with the literature, thresholds in the combined-masking condition were higher than would be predicted by an energy-sum of the effects of the individual forward and backward maskers. This is often referred to as ‘excess’ masking. The data were modeled by subjecting the amplitude of the stimuli to a power-law nonlinearity followed by a sliding temporal integrator (‘window’). It was assumed that threshold corresponds to a fixed signal-to-noise ratio at the output of the window. The best fits to the data were obtained using a power less than unity (0.5 to 0.7), i.e. by a compressive nonlinearity. Generally good fits to the data were achieved, indicating that the model is able to account for the decay of forward and backward masking as well as the effects of combining pairs of maskers (excess masking). The temporal windows derived from the data are also able to predict thresholds in decrement and increment detection tasks, and to account for the longer-term effects of masker duration in forward masking.

Vibrotactile temporal masking: Effects of multiple maskers

Previous investigations of temporal masking effects in vibrotactile detection tasks have generally produced results suggesting considerable similarities in stimulus processing between the tactile system and the auditory system. Auditory presentation of combinations of maskers (e.g., two forward maskers) has yielded masking effects in excess of the level of masking predicted by a simple intensity sum of each masker's individual effect. This "additional masking" has led to predictions of compressive nonlinearities in auditory stimulus processing. In the present study, two experiments were conducted to investigate further temporal masking phenomena for the tactile system. In the first experiment vibrotactile temporal masking functions for single forward and backward maskers were generated to examine trade-offs among values of masker duration, masking intensity, and interstimulus interval to maintain constant detectability of a target. Results suggested that stimuli in very close temporal proximity to the target might follow somewhat different trading equations from those farther removed. Overall, results were consistent with previous findings of temporal integration in the tactile system, and support the notion of an energy integrating mechanism. In the second experiment, pairs of maskers were presented to determine whether additional masking effects occurred in vibrotactile detection. In contrast to findings for auditory presentation, in which all masker configurations generated additional masking, the tactile results showed substantial amounts of additional masking only for pairs of backward maskers. Several possible explanations for this results are evaluated, and results are discussed in terms of similarities and dissimilarities in auditory and tactile temporal processing.

Improving the detectability of a brief tone in noise using forward and backward masker fringes: Monotic and dichotic presentations

A brief tonal signal simultaneously masked by a brief noise burst became easier to hear when the masker duration was increased. The signal was a 1000-Hz tone, 4 ms in duration; the masker was a wideband noise having a spectral notch 1400 Hz wide centered at 1000 Hz. Compared to performance with a 22-ms burst masker, average detection threshold across five subjects improved by 15 dB when a 150-ms masker "fringe" preceded the signal (forward fringe), and by 9 dB when the masker fringe followed the signal (backward fringe). Little improvement was observed in either condition when the fringe was presented to the ear contralateral to the signal/burst complex. However, when the fringe was presented to both ears and the signal/burst complex to just one ear, the forward fringe was about as helpful as when the stimuli were presented monotically, but the benefit of the backward fringe was substantially reduced. The backward-fringe advantage was restored by reducing the level, or delaying the onset, of the contralateral component of the fringe. The results suggest that the forward-fringe advantage is a robust phenomenon that is largely insensitive to input to the contralateral ear. In contrast, the backward-fringe advantage appears to be a fragile effect that can be affected by inputs from both ears.

The Effect of Forward Masker Duration, Rise/Fall Time, and Integrated Pressure on Auditory Brain Stem Evoked Responses in Human Newborns and Adults

The effects of duration, rise/fall time, integrated pressure, and intensity of a forward masker on the auditory brain stem evoked response (ABR) were examined in three studies. Keeping peak masker intensity constant, integrated masker pressure was a better predictor of ABR latency prolongation than masker duration or rise/fall time in human newborns. Masker effects varied as a function of masker intensity. Greater ABR latency prolongation was observed for more intense maskers. Furthermore, latency prolongation was observed with increasing masker duration for high-intensity, but not low-intensity maskers. Forward masking had a greater effect on newborn wave V latencies than adult wave V latencies; however, increasing masker duration had similar effects on newborn and adult wave V latencies.

Masker duration effects frequency resolution in simultaneous and nonsimultaneous masking

Frequency resolution was estimated as a function of masker duration in four masking conditions for three subjects. The signal was a 4-ms, 1000-Hz tone. The wideband masker (33-dB spectrum level) had a spectral notch of variable width centered on the signal. Masker duration was 6 (only in simultaneous masking), 62, or 412 ms. Rise/decay time was 2 ms for all stimuli. In two simultaneous-masking conditions, signal offset came 1 ms before masker offset (forward-fringe) or signal onset came 1 ms after masker onset (backward-fringe). In two nonsimultaneous-masking conditions, signal onset came immediately after masker offset (forward masking) or signal offset came immediately before masker onset (backward masking). In all conditions, threshold decreased as the width of the masker notch increased. However, the rate of the threshold decrease tended to increase with masker duration. Further, the threshold decrease was more marked in the two ‘‘forward’’ than in the two ‘‘backward’’ conditions. These data indicate that measured frequency resolution may dynamically change due to prior, or even subsequent, stimulation. [Work supported by the NIDCD.]

Effects of frequency and masker duration on free-field masking

The design of three-dimensional auditory displays for virtual environments should consider the potential for interference among sound sources. In the present study, free-field masking was measured between sound sources that were separated in azimuth or in elevation. The click train signal and noise masker were filtered to lie within low- (below 1.5 kHz), middle- (1.2–7.0 kHz), or high- (above 3.5 kHz) frequency regions. As expected, based on headphone data and on traditional models of binaural masking, there was a dramatic decrease in masking (as much as 18 kB) when the sound sources were separated within the horizontal plane. Similar decreases in masking (as much as 9 dB) were observed for the high-frequency signal and masker when they were separated within the median plane. Although phenomenologically similar, the masking release observed in the horizontal plane is likely to depend on interaural cues, whereas, these cues should play a minimal role within the median plane. Differences across frequency regions and differences between continuous and pulsed masker presentation are also discussed. [Work supported by AFOSR-91-0289.]

Behaviour of delayed evoked otoacoustic emission under forward masking paradigm.

The behaviour of delayed evoked otoacoustic emission (DEOAE) has been studied in normally hearing adults under a conventional forward-masking paradigm, and subjective measurements were carried out additionally for comparison. The clicks served as signals and the noise bursts were used as masker. In different experimental sets, signal and masker intensity, masker duration, and the interval between masker and maskee were altered. At masker levels corresponding to the subjective post-masking threshold of the clicks, the DEOAE was unaffected, i.e. had no noticeable alteration, compared with click stimulation without masking. Even at higher masker levels the inaudible clicks elicited clearly discernible DEOAEs. The forward-masking detection threshold of DEOAE ('DEOAE post-masking threshold') was reached at masker levels approximately 35 dB above the subjective post-masking threshold. The gap between subjective and DEOAE post-masking threshold vis-à-vis the masker level was also evident at different masker durations and different time intervals between masker and maskee. Central neural and peripheral receptor mechanisms are suggested to be the constituents of the masking phenomenon. The neural mechanism is involved at low masker levels. The receptor mechanism effectively joins the neural one at masker levels exceeding the threshold of psychoacoustic masking. The progressive increase in the number of auditory units from the periphery to the centre in the hearing system, linked with an increase in inhibition, can help to explain these effects.

Spectral integration of broadband signals in diotic and dichotic masking experiments

The method of Gässler [Acustica 4, 408-414 (1954)] was used to measure the audibility of multicomponent signals as a function of their bandwidth against a broadband, white-noise masker. Test signals were composed of 1 to 41 sinusoids with a spectral spacing of 10 Hz and were always spectrally centered around 400 Hz. Masker duration was 400 ms and the 300-ms signals were centered within the noise intervals. A three-interval forced-choice procedure with adaptive level adjustment was applied. NoSo, NoSm, NoS pi, and N pi So masked thresholds were obtained for four subjects. A comparison of the diotic and the three dichotic conditions yields no significant difference in the bandwidth dependence and suggests equal integration bandwidths for all conditions. However, the original results of Gässler could not be replicated: Neither were the overall levels of signals with a bandwidth below the critical bandwidth constant nor were the results for broadband signals in accordance with a single-band model of detection. The narrow-band data are much better described by calculating the overall signal level at the output of a rounded exponential filter [Patterson et al., J. Acoust. Soc. Am. 72, 1788-1803 (1982)] with an equivalent rectangular bandwidth of 65 Hz. For broader signal bandwidths, the signal level at threshold increases as predicted by a multiband model.

Evoked otoacoustic emission: behaviour under the forward masking paradigm.

The behaviour of evoked otoacoustic emission (EOAE) has been studied in normally hearing adults under the conventional forward masking paradigm. The clicks and tone pips served as signals, and the noise bursts were used as maskers. At noise burst levels, defined as psychoacoustic masking thresholds of signals (postmasking threshold), attenuation of EOAEs was just noticeable but did not occur in all records. The EOAEs were not completely eliminated even at noise levels exceeding the post-masking threshold by 30 dB. Central or neural and peripheral or receptor mechanisms are suggested to be the constituents of the masking phenomenon. The receptor mechanism effectively joins the neural one at masker levels, well exceeding the thresholds of psychoacoustic masking. The increase in masker duration and decrease in interval between masker and signal seem to accentuate the neural mechanism. As a result, the difference between masker levels, leading to psychoacoustic masking of the signal and having marked attenuating effects upon the EOAEs, respectively, is increased.

Forward pattern masking: effects of spatial frequency and contrast

Contrast thresholds of Gabor targets presented after sine-wave grating maskers were measured. All the patterns were isochromatic. The duration of both masker and target was 33 ms. The interstimulus interval was 33 ms. Masker spatial frequency, masker contrast, and target frequency were varied. Two experiments were conducted. In the first there were four target frequencies: 0.5, 1, 2, and 4 c/deg, with five masker frequencies covering a range of +/- 2 octaves around each and five masker contrasts (0.05-0.8). In the second there were three target frequencies (1.5, 2, and 2.83 cycles per degree), five masker frequencies around each, and 13 contrasts (0.005-0.8). Masking functions were bandpass, with maximum masking near the target frequency. Target thresholds increased with masker contrast. Over the shorter, sparser contrast range of experiment 1, functions appear roughly linear or concave upward. When they are plotted on log-log coordinates, functions for different masker frequencies diverge as masker contrast increases. Experiment 2 makes it apparent that some of these functions contain two increasing segments with a flat segment between them. These results are well described by a theory postulating that detection of these targets is mediated by two or three mechanisms that are differentially tuned to spatial frequency. The effect of the forward masker is briefly to desensitize the mechanisms that respond to it. The theory is fitted to the results of experiment 2, and its parameters are estimated.

Frequency-specific enhancement of the cochlear compound action potential: Influence of the forward masker

Under restricted frequency and intensity conditions, forward masking can result in the amplitude of the CAP being increased above its unmasked value (Henry, 1991). The present study provides a quantitative analysis of this enhancement effect. In response to forward maskers having the same frequency as the probe stimulus, central frequency (CF) enhancement varies as a function of the level of the forward masker: the lowest masker level at which it can reliably be detected is often well below the visual detection threshold of the CAP generated by the unmasked probe stimulus; the highest masker level at which it can reliably be detected corresponds to approximately 10 dB above the probe stimulus CAP threshold. A second low frequency (LF) enhancement region also exists, encompassing a narrow range of more intense maskers. CF enhancement can double the amplitude of the CAP, whereas LF enhancement is less pronounced. The magnitude of CF enhancement varies as a function of the duration of the forward masker, with longer durations generally increasing the magnitude of the effect. This duration effect, however, interacts with the level of the stimulus. Decreasing the interval between the end of the forward masker and the beginning of the probe increases the magnitude of CF enhancement.

The effect of masker spectral asymmetry on overshoot in simultaneous masking

"Overshoot" is a simultaneous masking phenomenon: Thresholds for short high-frequency tone bursts presented shortly after the onset of a broadband masker are raised compared to thresholds in the presence of a continuous masker. Overshoot for 2-ms bursts of a 5000-Hz test tone is described for four subjects as a function of the spectral composition and level of the masker. First, it was verified that overshoot is largely independent of masker duration. Second, overshoot was determined for a variety of 10-ms masker bursts composed of differently filtered uniform masking noise with an overall level of 60 dB SPL: unfiltered, high-pass (cutoff at 3700 Hz), low-pass (cutoff at 5700 Hz), and third-octave-band-(centered at 5000 Hz) filtered uniform masking noises presented separately or combined with different bandpass maskers (5700-16000 Hz, 5700-9500 Hz, 8400-16000 Hz) were used. Third, masked thresholds were measured for maskers composed of an upper or lower octave band adjacent to the third-octave-band masker as a function of the level of the octave band. All maskers containing components above the critical band of the test tone led to overshoot; no additional overshoot was produced by masker components below it. Typical values of overshoot were on the order of 12 dB. Overshoot saturated when masker levels were above 60 dB SPL for the upper octave-band masker. The standard neurophysiological explanation of overshoot accounts only partially for these data. Details that must be accommodated by any full explanation of overshoot are discussed.

Detection of silent temporal gaps in sinusoidal markers

Gap detection thresholds were measured by forced-choice procedure for conditions where the duration of a silent gap was varied adaptively between pairs of sinusoidal markers of the same or different frequency. Frequencies of the first sinusoid in a pair of markers ranged from F1 = 500 to 4000 Hz. Second-sinusoid marker frequencies F2 included F1 = F2, and usually frequencies 2%, 5%, 24%, and 50% higher than F1. In preliminary studies the role of presentation level (E/N0) on gap detection was considered. Preliminary data revealed confounding extraneous factors arising from gating transients and from overall stimulus (i.e., markers + gap) and/or masker duration cues. In the main experiments, the contributions of these extraneous cues were evaluated with experimental designs aimed at identifying and minimizing the confounding roles of these cues in gap detection. For conditions where extraneous gating transient cues were minimized (by presenting the sinusoidal markers in a continuous noise masker with random onset phase for the second sinusoid in every pair of markers) and overall stimulus duration cues were diminished (by randomizing the duration of each marker independently), gap detection thresholds increased from 5 to 90 ms as the frequency separation between F1 and F2 was increased by half an octave. When the gap detection thresholds were treated as filter attenuation values by normalizing and converting the data into decibels, the data were closely fit by the roex filter model. On average, the listeners' performances were modeled well by a constant-percentage (7%) bandwidth filter centered on F1.