Detection of partially filled gaps in noise and the temporal modulation transfer function

Results of experiments on the detection of silent intervals, or gaps, in broadband noise are reported for normal-hearing listeners. In some preliminary experiments, a gap threshold of about 2 ms was measured. This value was independent of the duration of the noise burst, variation of the noise level on each presentation, or the temporal position of the gap within the noise burst. In the main experiments, the thresholds for partial decrements in the noise waveform as well as brief increments were determined. As predicted by a model that assumes a single fixed peak-to-valley detection ratio, thresholds for increments are slightly higher than thresholds for decrements when the signal is measured as the change in rms noise level. A first-order model describes the temporal properties of the auditory system as a low-pass filter with a 7- to 8-ms time constant. Temporal modulation transfer functions were determined for the same subjects, and the estimated temporal parameters agreed well with those estimated from the gap detection data. More detailed modeling was carried out by simulating Viemeister's three-stage temporal model. Simulations, using an initial stage bandwidth of 4000 Hz and a 3-ms time constant for the low-pass filter, generate data that are very similar to those obtained from human subjects in both modulation and gap detection.     

Humans detect gaps in broadband noise according to effective gap duration without additional cues from abrupt envelope changes

Previous studies of behavior and IC single units in the mouse support theoretical expectations that gaps with ramped trailing markers have reduced detectability compared to equivalent gaps with ramped leading markers. In experiment 1, detection probability and response speeds of humans listening for gaps in broadband noise were investigated by independently varying either leading marker fall-time (FT) or trailing marker rise-time (RT). Gaps with silent duration of 1, 4, or 12 ms were presented 2 s into a 3-s noise burst, with either abrupt marker onsets and offsets or linearly ramped RT/FT of 2, 4, or 8 ms durations. Addition of a nonzero RT or FT to the gap silent period increased detectability and also increased reaction speed on trials with "Yes" response, but there was no difference in detectability or response speeds between RT and FT conditions. Experiment 2 extended this finding to gaps having two, one, or no abrupt marker edges. These findings suggest that human listeners do not make use of abrupt onset or offset information to enhance gap detection, but seem to rely on the effective sound level reduction associated with the gap for detection.     

Psychometric functions for gap detection in noise measured from young and aged subjects.

Psychometric functions for gap detection of temporal gaps in wideband noise were measured in a "yes/no" paradigm from normal-hearing young and aged subjects with closely matched audiograms. The effects of noise-burst duration, gap location, and uncertainty of gap location were tested. A typical psychometric function obtained in this study featured a steep slope, which was independent of most experimental conditions as well as age. However, gap thresholds were generally improved with increasing duration of the noise burst for both young and aged subjects. Gap location and uncertainty had no significant effects on the thresholds for the young subjects. For the aged subjects, whenever the gap was sufficiently away from the onset or offset of the noise burst, detectability was robust despite uncertainty about the gap location. Significant differences between young and aged subjects could be observed only when the gap was very close to the signal onset and offset.     

The effect of temporal placement on gap detectability

The detectability of a masked sinusoid increases as its onset approaches the temporal center of a masker. This study was designed to determine whether a similar change in detectability would occur for a silent gap as it was parametrically displaced from the onset of a noise burst. Gap thresholds were obtained for 13 subjects who completed five replications of each condition in 3 to 13 days. Six subjects were inexperienced listeners who ranged in age from 18 to 25 years; seven subjects were highly experienced and ranged in age from 20 to 78 years. The gaps were placed in 150-ms, 6-kHz, low-passed noise bursts presented at an overall level of 75 dB SPL; the bursts were digitally shaped at onset and offset with 10-ms cosine-squared rise-fall envelopes. The gated noise bursts were presented in a continuous, unfiltered, white noise floor attenuated to an overall level of 45 dB SPL. Gap onsets were parametrically delayed from the onset of the noise burst (defined as the first nonzero point on the waveform envelope) by 10, 11, 13, 15, 20, 40, 60, 110, 120, and 130 ms. Results of ANOVAs indicated that the mean gap thresholds were longer when the gaps were proximal to signal onset or offset and shorter when the gaps approached the temporal center of the noise burst. Also, the thresholds of the younger, highly experienced subjects were significantly shorter than those of the younger, inexperienced subjects, especially at placements close to signal onset or offset. The effect of replication (short-term practice) was not significant nor was the interaction between gap placement and replication. Post hoc comparisons indicated that the effect of gap placement resulted from significant decreases in gap detectability when the gap was placed close to stimulus onset and offset.     

Temporal gap resolution in narrow-band noises with center frequencies from 6000-14000 Hz

Temporal resolution was examined in normal-hearing subjects using a broadband noise and five narrow-band noises with center frequencies (fc) spaced 2 kHz apart between 6 and 14 kHz. Bandwidths of the narrow-band signals were equal to 0.16 fc, and broadband noise maskers with spectral notches were used to restrict the listening bands. Subjects used a Békésy procedure to track the minimum signal level required to keep a periodic temporal gap of fixed duration at threshold. Gap durations from 25 ms to the smallest trackable value were tested with each signal to generate performance curves, which showed the relationship between gap resolution and signal level in the low-to-moderate intensity range. Results showed that gap resolution improved progressively with increased signal level to about 35 dB SL, where minimum gap thresholds of about 3 ms were observed for all signals. These results, when combined with previous low-frequency data, indicate that gap threshold decreases systematically with increased signal frequency to about 5 kHz, and asymptotes at 2-3 ms for higher frequencies. In the context of functional models, the frequency effect is qualitatively consistent with the notion that both the auditory filter and a sensory integrator operate in series to govern temporal resolution in audition.     

Gap detection for similar and dissimilar gap markers

Detection thresholds for temporal gaps between markers of dissimilar frequency are usually elevated with respect to thresholds for gaps between markers of similar frequency. Because gaps between markers of dissimilar frequency represent both a spectrally based perceptual discontinuity as well as a temporal discontinuity, it is not clear what factors underlie the threshold elevation. This study sought to examine the effects of perceptual dissimilarities on gap detection. The first experiment measured gap detection for configurations of narrow-band gap markers comprised of pure tones, frequency-modulated tones, and amplitude-modulated tones. The results showed that gap thresholds for frequency-disparate pure-tone markers were elevated with respect to isofrequency tonal markers, but that perceptual discontinuities between markers restricted to the same frequency region did not uniformly elevate threshold. The second experiment measured gap detection for configurations of markers where the leading and trailing markers could differ along the dimensions of bandwidth, duration, and pitch. The results showed that, in most cases, gap detection deteriorated when the bandwidth of the two markers differed, even when the spectral content of the narrower-band marker was completely subsumed by the spectral content of the wider-band marker. This finding suggests that gap detection is sensitive to spectral dissimilarity between markers in addition to spectral discontinuity. The effects of marker duration depended on the marker bandwidth. Pitch differences across spectrally similar markers had no effect.     

Modulation and gap detection for broadband and filtered noise signals

Modulation detection thresholds (as a function of sinusoidal amplitude modulation frequency) and temporal gap detection thresholds were measured for three low-pass-filtered noise signals (fc = 1000, 2000, and 4000 Hz), a high-pass-filtered noise signal (fc = 4000 Hz), and a broadband signal. The two latter noise signals were effectively low-pass filtered (fc = 6500 Hz) by the earphone. Each of the filtered signals was presented with a complementary filtered noise masker. Modulation and gap detection thresholds were lowest for the broadband and high-pass signals. Thresholds were significantly higher for the low-pass signals than for the broadband and high-pass signals. For these tasks and conditions, the high-frequency content of the noise signal was more important than was the signal bandwidth. Sensitivity (s) and time constant (tau) indices were derived from functions fitted to the modulation detection data. These indices were compared with gap detection thresholds for corresponding signals. The gap detection thresholds were correlated inversely (rho = -1.0, p less than 0.05) with s (i.e., smaller gap detection thresholds were correlated with greater sensitivity to modulation), but were not correlated significantly with tau, which was relatively invariant across signal conditions.     

Spectro-temporal integration in signal detection

This paper is concerned with aspects of temporal integration and across-frequency integration in signal detection. Previous experiments on the detection of brief broadband signals (clicks) in continuous broadband noise revealed efficient spectral integration. The extent to which this effect is restricted to a critical time window was investigated by manipulating the temporal relations among the signal components in different frequency regions. In a typical experiment, the signal consists of nine brief Gaussian-shaped tone pulses, equally distributed at 1/3-oct intervals, each with a spectral width of about 1/3 oct, and each equally detectable in white noise. In the synchronized condition (i.e. coinciding peaks of the nine Gaussian envelopes), the detection threshold is reached when the levels of the nine individual tone pulses are about 8 dB below their individual threshold levels (efficient spectral integration). When the signal is progressively desynchronized (i.e. noncoinciding peaks of the Gaussian envelopes), detection threshold is found to increase. This suggests that efficient spectral integration in signal detection is confined to a narrow time window, with a typical value of 30 ms. Similar experiments were performed with respect to the efficiency of temporal integration. For constant-duration signals (100 ms), the detection threshold is found to increase when progressively widening signal bandwidth. The data indicate that the efficient temporal integration in signal detection is confined to a narrow frequency window, which, not surprisingly, corresponds to the critical bandwidth.     

Vibrotactile temporal gap detection as a function of age

The ability of subjects to detect temporal gaps between bursts of sinusoids or bursts of bandlimited noise was measured to evaluate the phenomenon of tactile "sensory persistence" in older persons. Vibratory stimuli were delivered to the right thenar eminence of 27 subjects ranging in age from 8-75 years. The subjects' task was to detect the presence of a silent interval or "gap" between flanking 350-ms vibrotactile stimuli. The gap-detection threshold, expressed as the amplitude of vibration relative to the absolute detection threshold, decreased as the gap duration increased and was higher for gaps in noise than for gaps in sinusoids. The threshold for detecting short gaps increased with age for noise stimuli, but not for sinusoidal stimuli. Furthermore, the gap-detection threshold recovered more rapidly in older subjects for noise stimuli, but less rapidly in older subjects for sinusoidal stimuli. Because of these differences, it appears that the effects of age on gap detection cannot be due to a simple increase in sensory persistence, but may be due to multiple processes.     

Spectral differences in the ability of temporal gaps to reset the mechanisms underlying overshoot

When very brief tonal signals are presented immediately after the onset of a gated noise masker, detectability can be 10-20 dB worse than when the signal is delayed by several hundred milliseconds, an effect known as the overshoot. It has long been known that, when an "onset" is created in an otherwise continuous, broadband masker by briefly turning it off and on again, the detectability of a brief signal presented soon after this temporal gap will decline gradually as the gap is increased from a few milliseconds to a few hundred milliseconds. In other words, the auditory system recovers to its quiescent, resting state following an adequate silent interval. Here, the broadband maskers consisted of three adjacent spectral bands--one centered on the frequency of the tonal signal, one low passed below the lower edge of the center band, and one high passed above the upper edge of the center band. The signal was a 2500-Hz tone having a total duration of 6 ms. In different blocks of trials, either all three bands, only the center band, or only the two flanking bands were temporally gapped by a duration ranging from 10-300 ms. When the center band was about 750 Hz wide (about 2.5 critical bandwidths), this differential gapping process resulted in typical recovery functions when all three bands (the entire spectrum) or when just the two flanking bands were gapped. However, when only the center band was gapped, there was no evident recovery--rather, detectability remained near the signal level required with a continuous masker, even for a gap duration of 300 ms.(ABSTRACT TRUNCATED AT 250 WORDS)     

Detection of temporal gaps in sinusoids by normally hearing and hearing-impaired subjects

A two-alternative forced-choice task was used to measure psychometric functions for the detection of temporal gaps in a 1-kHz, 400-ms sinusoidal signal. The signal always started and finished at a positive-going zero crossing, and the gap duration was varied from 0.5 to 6.0 ms in 0.5-ms steps. The signal level was 80 dB SPL, and a spectrally shaped noise was used to mask splatter associated with the abrupt onset and offset of the signal. Two subjects with normal hearing, two subjects with unilateral cochlear hearing loss, and two subjects with bilateral cochlear hearing loss were tested. The impaired ears had confirmed reductions in frequency selectivity at 1 kHz. For the normal ears, the psychometric functions were nonmonotonic, showing minima for gap durations corresponding to integer multiples of the signal period (n ms, where n is a positive integer) and maxima for durations corresponding to (n - 0.5) ms. For the impaired ears, the psychometric functions showed only small (nonsignificant) nonmonotonicities. Performance overall was slightly worse for the impaired than for the normal ears. The main features of the results could be accounted for using a model consisting of a bandpass filter (the auditory filter), a square-law device, and a sliding temporal integrator. Consistent with the data, the model demonstrates that, although a broader auditory filter has a faster transient response, this does not necessarily lead to improved performance in a gap detection task. The model also indicates that gap thresholds do not provide a direct measure of temporal resolution, since they depend at least partly on intensity resolution.     

Detection of temporal gaps in bandlimited noise: effects of variations in bandwidth and signal-to-masker ratio

Thresholds were measured for the detection of a temporal gap in a bandlimited noise signal presented in a continuous wideband masker, using an adaptive forced-choice procedure. In experiment I the ratio of signal spectrum level to masker spectrum level (the SMR) was fixed at 10 dB and gap thresholds were measured as a function of signal bandwidth at three center frequencies: 0.4, 1.0, and 6.5 kHz. Performance improved with increasing bandwidth and increasing center frequency. For a subset of conditions, gap threshold was also measured as bandwidth was varied keeping the upper cutoff frequency of the signal constant. In this case the variation of gap threshold with bandwidth was more gradual, suggesting that subjects detect the gap using primarily the highest frequency region available in the signal. At low center frequencies, however, subjects may have a limited ability to combine information in different frequency regions. In experiment II gap thresholds were measured as a function of SMR for several signal bandwidths at each of three center frequencies: 0.5, 1.0, and 6.5 kHz. Gap thresholds improved with increasing SMR, but the improvement was minimal for SMRs greater than 12-15 dB. The results are used to evaluate the relative importance of factors influencing gap threshold.     

Spectral shape discrimination of narrow-band sounds.

Measurements are reported on the detectability of signals added to narrow-band sounds. The narrow-band sounds had a bandwidth of 20 Hz and were either Gaussian noise with flat amplitude spectra or sets of equal-amplitude sinusoidal components whose phases were chosen at random. Four different kinds of sinusoidal signals were used. Two signals produced symmetric changes in the audio spectrum adding a component either at the center of the spectrum or at both ends. The other two signals produced asymmetric changes adding a component at either end of the spectrum. The overall level of the sound was randomly varied on each presentation, so that the presence of a signal was largely unrelated to the absolute level of the signal component(s). A model is proposed that assumes the detection of the symmetric signals is based on changes in the shape of the power spectrum of the envelope. Such changes in the envelope power spectrum are probably heard as changes in the "roughness" or "smoothness" of the narrow-band sound. The predictions of this model were obtained from computer simulations. For the asymmetric signals, the most probable detection cues were changes in the pitch of the narrow-band sound. Results from a variety of different experiments using three listeners support these conjectures.     

Spectral integration of broadband signals in diotic and dichotic masking experiments.

The method of G?ssler [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 G?ssler 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.     

Level-dependent changes in detection of temporal gaps in noise markers by adults with normal and impaired hearing

Compression in the basilar-membrane input-output response flattens the temporal envelope of a fluctuating signal when more gain is applied to lower level than higher level temporal components. As a result, level-dependent changes in gap detection for signals with different depths of envelope fluctuation and for subjects with normal and impaired hearing may reveal effects of compression. To test these assumptions, gap detection with and without a broadband noise was measured with 1,?000-Hz-wide (flatter) and 50-Hz-wide (fluctuating) noise markers as a function of marker level. As marker level increased, background level also increased, maintaining a fixed acoustic signal-to-noise ratio (SNR) to minimize sensation-level effects on gap detection. Significant level-dependent changes in gap detection were observed, consistent with effects of cochlear compression. For the flatter marker, gap detection that declines with increases in level up to mid levels and improves with further increases in level may be explained by an effective flattening of the temporal envelope at mid levels, where compression effects are expected to be strongest. A flatter effective temporal envelope corresponds to a reduced effective SNR. The effects of a reduction in compression (resulting in larger effective SNRs) may contribute to better-than-normal gap detection observed for some hearing-impaired listeners.     

Temporal decline of masking and comodulation detection differences

Comodulation detection differences (CDDs) were studied using flanking bands that were either gated simultaneously with the signal band (burst) or gated at varying times prior to signal onset (fringed). Used for these experiments were a signal band centered at 1250 Hz and four flanking bands centered at 450, 850, 1650, and 2050 Hz; all bands were 100 Hz wide. In different conditions, the temporal envelope of the signal band was either the same as (correlated), or different from (uncorrelated), the common envelope of the four flanking bands, or the temporal envelopes of all of the bands were different (all-uncorrelated). For 8 of the 13 listeners, signal detectability improved by as much as 25 dB as the temporal fringe of the flanking bands was increased from 5 to about 700 ms. This temporal decline of masking was similar, but not identical, for the correlated, uncorrelated, and all-uncorrelated conditions. Results of this sort are reminiscent of several related findings that have been attributed to auditory adaptation or enhancement, or to a temporally developing critical-band filter. The other 5 of the 13 listeners were generally more sensitive than the majority, and they showed little or no improvement in detectability as fringe duration was varied. Large individual differences of this sort are not uncommon in the adaptation and comodulation literatures. As signal duration was changed from 50 to 240 ms, temporal integration was less in the correlated condition than in the uncorrelated condition, thereby producing a larger CDD with the longer signal. When the fringe followed the observation interval instead of preceding it, the results were equivocal because detectability improved for the majority of subjects and worsened for the minority. In follow-up experiments, different subsets of these four flanking bands were used. When temporal gaps of varying duration were inserted into the flanking band(s) immediately prior to the observation intervals, it was found that a temporal gap as long as 355 ms was not sufficient to reset the mechanisms underlying the temporal decline of masking.     

Interaural correlation discrimination: II. Relation to binaural unmasking

Many theoretical models of binaural interaction assume that sensitivity to interaural correlation underlies binaural unmasking. This paper explores the extent to which sensitivity to changes in interaural correlation implied by results from binaural detection experiments are consistent with sensitivity to changes in interaural correlation implied by results from binaural detection experiments are consistent with sensitivity to changes in interaural correlation measured directly in correlation discrimination experiments. The vehicle for this exploration is a simplified model of the underlying processes assumed by many models of binaural unmasking for the detection of narrow-band signals in the presence of broadband noise. Consideration is given to psychometric function slopes, detection thresholds, bandwidth effects, duration effects, level effects, and interaural-parameter effects. Although many of the results obtained from our analysis are consistent with the notion that the cue in binaural detection tasks is a change in interaural correlation, some significant inconsistencies are noted.     

Discrimination of interaural envelope correlation and its relation to binaural unmasking at high frequencies.

Listeners' sensitivity to interaural correlation of the envelope of high-frequency waveforms and whether such sensitivity might account for detectability in a masking-level difference paradigm were assessed. Thresholds of interaural envelope decorrelation (from a reference correlation of 1.0) were measured for bands of noise centered at 4 kHz and bandwidths ranging from 50-1600 Hz. Decorrelation of the envelope was achieved by "mixing" two independent narrow-band noises. Separately, with the same listeners, NoSo and NoS pi detection thresholds were measured for maskers of the same center frequency and bandwidths. For bandwidths of noise up to about 400 Hz, listeners were similarly sensitive to interaural decorrelation in both types of task. However, for bandwidths greater than 400 Hz or so, while sensitivity in the discrimination task was unaffected, sensitivity was reduced in the NoS pi conditions. Additional data suggested that listeners were able to maintain their sensitivity independent of bandwidth in the discrimination task by focusing on binaural information within select spectral regions of the stimuli.     

Profile analysis and background noise

Spectral shape discrimination, or profile analysis, of complex waveforms (21 components) in the presence of broadband noise and special sinusoidal maskers of random amplitude was studied. The first experiment involved the discrimination between a standard flat spectrum and a "rippled" spectrum in broadband noise of different spectrum levels. Thresholds obtained under control conditions, without noise or without a standard, were used to estimate constants of an equation that predict thresholds where standard and noise are both present. The model assumes an external variance, produced by the noise, is added linearly to an internal variance caused by the flat standard. The mean squared error is less than 2 dB. The second experiment involved the detection of an increment on the center component of the 21-component standard. Added to the standard was an additional masking sinusoid of random amplitude. Both the frequency and the range of the random amplitude were varied and both showed a systematic influence on the detectability of the 1000-Hz increment.     

A binaural analog of gap detection.

The temporal resolution of the binaural auditory system was measured using a binaural analog of gap detection. A binaural "gap" was defined as a burst of interaurally uncorrelated noise (Nu) placed between two bursts of interaurally correlated noise (N0). The Nu burst creates a dip in the output of a binaural temporal window integrating interaural correlation, analogous to the dip created by a silent gap in the output of a monaural temporal window integrating intensity. The equivalent rectangular duration (ERD) of the binaural window was used as an index of binaural temporal resolution. In order to derive the ERD, both the shortest-detectable binaural gap and the jnd for a reduction in interaural correlation from unity were measured. In experiment 1, binaural-gap thresholds were measured using narrow-band noise carriers as a function of center frequency from 250 to 2000 Hz (fixed 100-Hz bandwidth) and a function of lower-cutoff frequency from 100 to 400 Hz (fixed 500-Hz upper-cutoff frequency). Binaural-gap thresholds (1) increased significantly with increasing frequency in both tasks, and (2) at frequencies below 500 Hz, were shorter than corresponding silent-gap thresholds measured with the same N0 noises. In experiment 2, interaural-correlation jnd's were measured for the same conditions. The jnd's also increased significantly with increasing frequency. The results were analyzed using a temporal window integrating the output of a computational model of binaural processing. The ERD of the window varied widely across listeners, with a mean value of 140 ms, and did not significantly depend on frequency. This duration is about an order of magnitude longer than the ERD of the monaural temporal window and is, therefore, consistent with "binaural sluggishness."