Broadband Noise Masker Research Articles

A model of the effective auditory signal processing. I: Masking experiments

A model is described that simulates the detection characteristics of the human auditory system for different psychoacoustical tasks. An optimal detection process is attached to the model after the preprocessing stages. The statistical basis was provided by the assumption of an internal Gaussian noise following nonlinear processing in the peripheral auditory system. The nonlinear processing describes essential features of human temporal processing, such as forward masking, temporal integration, and dynamic range compression. Simulations and measurements of backward, simultaneous, and forward masking are presented employing deterministic (frozen) and stochastic (running) broadband noise maskers. The stimuli were presented to the model in an paradigm (adaptive forced-choice procedure) identical to the one used with human subjects. In most cases, a very good agreement between simulated and measured thresholds is achieved.

Masking of tones and formant frequencies

The purpose of this study was to find out whether results of auditory masking theory, established primarily for pure tones, can be used to predict masking of vowel formant frequencies by broadband noise maskers. For this purpose, three sets of 2AFC experiments were conducted. In each experiment, four frozen white-noise waveforms (with similar statistical properties) were used. The task in the first set of experiments, was tone-in-noise detection in which the tones were located at 1100 and 1800 Hz. In the second set, one-formant synthetic vowels were used in vowel-in-noise detection tasks. The formant was located at 1100 Hz in one experiment, and at 1800 Hz, in another. Finally, listeners were presented with pairs of four-formant synthetic vowels in noise: one pair had identical vowels and the second vowel in the second pair had a missing second formant frequency (F2). Subjects were instructed to choose the interval thought to contain the different pair of vowels. The four-formant synthetic vowel was either /ɑ /-like (formants at 625, 1100, 2700, and 3250 Hz) or /schwa/-like (formants at 500, 1800, 2700, and 3500 Hz). It was found that, on the average, the masked thresholds of tones and of formant frequencies were predicted well from the signal-to-noise ratio in a critical band centered around the tone or the formant frequency. [Work supported by NIH and NSF.]

The effect of pure-tone forward masking on overshoot

The overshoot effect can be reduced by temporary hearing loss induced by aspirin or exposure to intense sound. The present study simulated a hearing loss at 4.0 kHz via pure-tone forward masking and examined the effect of the simulation on threshold for a 10-ms, 4.0-kHz signal presented 1 ms after the onset of a 400-ms, broadband noise masker whose spectrum level was 20 dB SPL. Masker frequency was 3.6, 4.0, or 4.2 kHz, and masker level was 80 dB SPL. Subject-dependent delays were determined such that 10 or 20 dB of masking at 4.0 kHz was produced. In general, the pure-tone forward masker did not reduce the simultaneous-masked threshold, suggesting that elevating threshold with a pure-tone forward masker does not sufficiently simulate the effect of a temporary hearing loss on overshoot.

Ipsilateral, contralateral, and binaural masking effects on the human brain-stem auditory-evoked responses to click stimuli

Two experiments concerning the effects of continuous broadband noise on wave V of the click-evoked human BAER are reported. Experiment I compared the effects of broadband masking noise presented ipsilaterally, contralaterally, and biaurally on the BAER to 100-dB pSPL monaurally presented clicks. For noise levels up to 75 dB SPL, contralateral masking had no effect on either the latency or the amplitude of wave V. Ipsilateral and binaural noise levels above 35-45 dB SPL increased wave V latency and decreased its amplitude, and the magnitudes of these effects were similar for ipsilateral and binaural noise conditions. Experiment II compared the effects of broadband masking noise on wave V to 100-dB pSPL clicks presented monaurally to each ear, and binaurally, with the noise being presented to the ear(s) of click presentation. In agreement with experiment I, noise levels above 35-45 dB SPL increased wave V latency and decreased wave V amplitude. There were no significant differences for right versus left ear. Wave V latency did not vary significantly for monaural versus binaural presentation, while wave V amplitude was larger for binaural than monaural presentation modes for all noise-level conditions.

Overshoot in subjects with sensorineural hearing impairment

Overshoot was measured in both ears in subjects with normal hearing and in subjects with unilateral high‐frequency, sensorineural hearing loss by measuring threshold for a 10‐ms signal presented 1 or 195 ms after the onset of a 400‐ms broadband noise masker. Masker spectrum level was 20, 30, or 40 dB SPL. At 4.0 kHz, where absolute thresholds were elevated in the impaired ears, overshoot (the difference between thresholds at 1 and 195 ms) was small or absent. This may indicate that hearing loss alters a mechanism contributing to overshoot, or it may reflect the fact that sufficient amounts of masking (greater than about 20 dB) could not be produced except at high masker levels, where overshoot can be small or absent in normal ears [S. P. Bacon, J. Acoust. Soc. Am. (in press)]. In addition, overshoot was smaller at 1.0 kHz in the impaired ear than in the normal ear of the unilaterally impaired subjects, despite the fact that absolute thresholds were normal at that frequency. This, too, may reflect a damaged peripheral mechanism, or it may reflect the normal consequence of a relatively narrow, “effective” masker bandwidth. [Work supported by NIDCD.]

On the frequency regions governing overshoot

Detectability of a brief, high‐frequency tone can improve as the onset of that tone is delayed relative to the onset of a long‐duration, broadband noise masker. This effect, known as overshoot, is, in part, consistent with adaptation at the auditory nerve, although recent psychophysical experiments suggest that other mechanisms also may be involved. To gain more insight into the mechanisms underlying overshoot, several experiments were designed primarily to manipulate the spectral properties of a 200‐ms noise precursor (N0 = 20 dB SPL) whose offset preceded the onset of a 400‐ms noise masker (N0 = 20 dB SPL) by 1 ms. Depending upon its spectral composition, the precursor could reduce or eliminate overshoot by reducing threshold near masker onset. The assumption here is that the frequency regions of the precursor that reduce overshoot are those that are, in turn, responsible for overshoot. The results indicate that overshoot is governed by a relatively large frequency region that is clearly not confined to the 4.0‐kHz signal frequency region, and that frequency regions above the signal are more important for overshoot than are frequency regions below the signal. [Work supported by NIDCD.]

Stimulus dependencies of the gerbil brain-stem auditory-evoked response (BAER). III: Additivity of click level and rate with noise level

Two experiments were performed that evaluated the effects of ipsilateral-direct broadband noise maskers on the gerbil brain-stem auditory-evoked response (BAER) to click stimuli. In experiment 1, clicks were presented at 27 Hz at levels including 70, 80, 90, and 100 dB pSPL. Noise conditions included a no-noise control, and included noise levels varying in 10-dB increments from 20 dB SPL to a maximum noise level of 50, 60, 70, and 80 dB SPL for click levels of 70, 80, 90, and 100 dB pSPL, respectively. Gerbil BAER peaks were labeled with small roman numerals to distinguish them from human BAER peaks. The dependent variables included waves i and v latencies and amplitudes. Peak latencies increased and peak amplitudes decreased with decreasing click level and increasing noise level. To a first approximation, peak latencies and amplitudes showed changes with increasing noise level that were similar across click level. With increasing click level, there was little or no effect on the i-v interval. There was an increase in the i-v interval with increasing noise level. In experiment 2, click level was held constant at 90 dB pSPL, and click rates included 15, 40, 65, and 90 Hz. For each click rate, noise conditions included a no-noise control, and noise levels included 20, 30, 40, 50, 60, and 70 dB SPL. With increasing click rate and noise level, there was an increase in peak latencies, an increase in the i-v interval, and a decrease in peak amplitudes. The magnitude of peak latency and amplitude shifts with increasing click rate was dependent on noise level. Specifically, the magnitude of rate-dependent changes decreased with increasing level of broadband noise. These data are compared to human BAER experiments, and are found to be in fundamental agreement.

The effect of interaural phase on frequency discrimination with broad- and narrow-band maskers

With both broad-band and narrow-band noise maskers there is a small range of signal-to-noise ratios where, with the maskers in-phase and signals 180° out-of-phase at the ears, detection performance grows with increasing signal-to-noise ratio but frequency discrimination performance does not. The observation may have significant consequences for our understanding of the way in which frequency is coded at low signal-to-noise ratios.

Comparison of the noise attenuation of three audiometric earphones, with additional data on masking near threshold

The noise-excluding properties of a standard supra-aural audiometric earphone, a widely used circumaural-supra-aural combination, and an insert earphone sealed to the ear with a vinyl foam eartip were measured in a diffuse-field room complying with ANSI S12.6-1984. Data on attenuation were obtained monaurally with the nontest ear plugged and muffed. Results for the supra-aural earphones generally agreed well with previously reported measurements. A broadband masking noise was used to directly test the ANSI S3.1-1977 permissible background noise levels for measuring to audiometric zero using standard audiometric earphones. This "ANSI noise" raised the average thresholds of 15 normal-hearing test subjects by 3 to 5 dB at the octave frequencies from 500 to 4000 Hz. With a noise conforming to the less stringent OSHA-1983 regulation, average thresholds were elevated 9 to 17 dB. An "ENT office noise" with an overall sound level of 54 dBA raised average thresholds even further, by as much as 29 dB at 500 Hz. Use of the circumaural system in the office noise limited the threshold elevation to 11, 5, 2, and 0 dB at the four octave frequencies tested. With the fully ("deeply") inserted foam eartips, the threshold elevation in the simulated office noise was 2 dB or less at all test frequencies. Actual threshold elevations agreed closely with predictions based on a critical ratio calculation utilizing measured sound field noise levels and measured earphone attenuation values.

Stimulus dependencies of the gerbil brain-stem auditory-evoked response (BAER). II: Effects of broadband noise level and high-pass masker cutoff frequency across click polarity.

Two experiments concerning the effects of masking noise on the gerbil brain-stem auditory-evoked response (BAER) are reported. Experiment 1 evaluated the effects of broadband masking noise on the BAER obtained to condensation and rarefaction clicks. With increasing noise level, there was an increase in BAER peak latencies, an increase in the i-v interval, and a decrease in peak amplitudes. Experiment 2 evaluated the effects of high-pass masking noise on the BAER obtained to condensation and rarefaction clicks. Both high-pass responses and derived-band responses were evaluated. For high-pass responses, with decreasing masker cutoff frequency, there was an increase in BAER peak latencies, a decrease in the i-v interval, and a decrease in peak amplitudes. For derived-band responses, with decreasing derived-band frequency, there was an increase in peak latencies and a decrease in the i-v interval. A comparison of wave i and wave v amplitudes across derived-band frequency demonstrates a greater contribution of high-frequency cochlear regions to wave i than wave v. Small, insignificant, effects of click polarity on BAER peak amplitudes were observed. These trends were in the direction seen in a companion paper [R. Burkard and H. F. Voigt, J. Acoust. Soc. Am. 85, 2514-2525 (1989)] and were, in general, reduced by the presence of broadband or high-pass maskers.

Temporal integration in an anuran auditory nerve

We determined temporal integration in individual auditory nerve fibers of the arboreal frog, Eleutherodactylus coqui by measuring the change in rate threshold to tonal stimulation for tone burst durations from 20 to 450 ms. Temporal integration was quantified in two ways: (1) we calculated the temporal integration time constant of the fiber (τ) and (2) we calculated the shift in threshold per decade increment of the tone stimulus (dB/decade). In some cases, the procedure was repeated using a continuous, broadband noise masker (20–50 dB/Hz). Our results indicate that low frequency fibers (CF < 0.50 kHz) have the longest mean integration time (274 ms), mid-frequency fibers (0.50 to 1.30 kHz) have the shortest mean integration time (183 ms) and high frequency fibers (CF > 1.3 kHz) have an intermediate mean integration time (235 ms). Continuous noise increased the integration times of some, but not all fibers and caused some fibers which did not display temporal integration to do so. We investigated the possibility that these changes may be caused by a decrease in the slope of the rate-intensity function (measured between + 5 and + 15 dB re threshold) with the addition of the continuous noise masker. The slope of the rate-intensity function decreased (from 73 spikes/s/dB to 49 spikes/s/dB) with the addition of the continuous noise for those fibers ( N = 28) that showed an increase in temporal integration with the addition of noise. However, the slopes of the rate-intensity functions also decreased by 30% for those fibers ( N = 8) that did not show increasing temporal integration.

Masking patterns in the bullfrog (Rana catesbeiana). II: Physiological effects.

Responses of individual eighth-nerve fibers in the bullfrog (Rana catesbeiana) were measured to tone bursts at best frequency against a background of continuous, broadband masking noise. These data were used to calculate critical masking ratios to describe the fibers' responses to tones embedded in noise. In the frequency response range of the amphibian papilla (100-1000 Hz), critical ratios increase with tone frequency. Critical ratios of basilar papilla fibers (1000-2000 Hz) are generally higher than those of amphibian papilla fibers. Critical ratios are also significantly related to fiber threshold such that fibers with high thresholds, regardless of their best frequencies, have higher critical ratios and are thus less selective to signals embedded in noise. Critical ratios based on neural responses show a somewhat different frequency-dependent trend than do critical ratios based on psychophysical data presented previously for this species [A. M. Simmons, J. Acoust. Soc. Am. 83, 1087-1092 (1988a)]. In addition, these neural critical ratios do not appear to be level independent, as are psychophysical critical ratios. The data suggest that frequency selectivity of hearing in the bullfrog as measured behaviorally is probably not mediated solely by spectral filtering in the auditory periphery.

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.]

Absence of overshoot in a dichotic masking condition

Brief tonal signals presented soon after the onset of a masking noise are known to be less detectable than signals delayed by several hundred milliseconds. This difference in detectability is known as the "overshoot." Signals of two sorts were studied here--either interaurally in phase (S o) or interaurally out of phase by 180 degrees (S pi). When S omicron signals of 750 Hz and about 14 ms in duration were presented 4 ms after the onset of a diotic, broadband masking noise (N o), detectability was about 6 dB worse than when the signal was presented 325 ms after onset. By contrast, there was no such overshoot when S pi signals were presented at varying times after masker onset; detectability was about the same for all values of signal delay. Accordingly, the difference in performance between N o S o and N o S pi--the masking-level difference or MLD--was large (about 16 dB) with the shortest delays used and diminished (to about 9 dB) as the delay was increased. This absence of overshoot with the S pi signals is in accord with the well-established view that detectability in the dichotic masking conditions is based upon different stimulus information from that used in the diotic masking conditions. Specifically, the evidence confirms the common view that detectability in the diotic conditions is based more or less directly on neural firing rate, whereas, in the dichotic conditions, it is based upon interaural time differences encoded in the periodicity of neural firings.

Binaural unmasking in infants.

Localization responses to a broadband noise signal presented against a broadband noise masker were obtained from 12-month-old infants and adults. Two loudspeakers, one to the left and one to the right of the listener, continuously presented identical broadband maskers. On a trial, a broadband signal was added to one of the loudspeakers. Subjects were required to identify the loudspeaker producing the signal. Noise signals were either coherent (from the same noise generator) or incoherent (from an independent noise generator). Both infants and adults found it easier to locate the incoherent signals even when the two types of signals were adjusted to produce equal increments in power. Since monaural performance, after this adjustment, should be equivalent for the two cases, superior performance for incoherent signals implies that binaural processing is involved. The same result was observed in control experiments in which coherent and incoherent signals were presented over earphones to adults. These results suggest that the mechanisms responsible for binaural unmasking are operative by 12 months of age.

Masker-bandwidth dependence in homophasic and antiphasic tone detection.

Thresholds for the detection of 500-ms tones in noise at either 0.25 or 4 kHz and either interaurally in-phase (NoSo) or out-of-phase (NoS pi) were measured as a function of the bandwidth of a diotic masking noise (with total noise power held constant). NoSo thresholds followed the classic trend indicative of an effective critical band in noise masking. NoS pi thresholds also indicated critical-band filtering, but with a wider effective critical bandwidth. At subcritical masker bandwidths, for both 0.25 and 4 kHz, NoS pi thresholds increased with an increase in noise bandwidth, despite the fact that total noise power was constant. This latter finding is attributed to binaural insensitivity to rapid fluctuations in the interaural cues that subserve detection in the NoS pi condition. These results suggest a new interpretation for the small difference between NoSo and NoS pi thresholds measured with high-frequency tones in a broadband noise masker. This interpretation is based partly on the inability to utilize rapidly varying interaural cues, partly on out-of-band interference effects, and partly on loss of information related to stimulus fine structure.

Frequency selectivity of hearing in the green treefrog,Hyla cinerea

Frequency selectivity of hearing was measured in the green treefrog, Hyla cinerea. A psychophysical technique based on reflex modification was used to obtain masked threshold estimates for pure tones (300-5,400 Hz) presented against two levels of broadband masking noise. A pure tone (S-1) presented 200 ms prior to a reflex-eliciting stimulus (S-2) inhibited the motor reflex response to S-2. The magnitude of this reflex modification effect varied systematically with the sound pressure level (SPL) of S-1, and threshold was defined as the SPL of S-1 at which the reflex modification effect disappeared. Masked thresholds were used to calculate critical ratios, an index of the auditory system's frequency selectivity. The frequency selectivity of the treefrog's hearing is greatest and critical ratios are lowest (22-24 dB) at about 900 and 3,000 Hz, the two spectral regions dominant in the male treefrog's species-specific advertisement call. These results suggest that the treefrog's auditory system may be specialized to reject noise at biologically-relevant frequencies. As in other vertebrates, critical ratios remain constant when background noise level is varied; however, the shape of the treefrog's critical ratio function across frequencies differs from the typical vertebrate function that increases with increasing frequency at a slope of about 3 dB/octave. Instead, the treefrog's critical ratio function resembles its pure tone audiogram. Although the shape of the treefrog's critical ratio function is atypical, the critical ratio values themselves are comparable to those of many other vertebrates in the same frequency range. Critical ratio values here measured behaviorally do not match critical ratio values previously measured physiologically in single eighth nerve fibers.(ABSTRACT TRUNCATED AT 250 WORDS)

Frequency effects in backward masking.

Frequency effects in auditory backward masking were examined by psychophysically determining the thresholds of 0.5-, 1-, 3-, 5-, and 7-kHz sinusoidal probes followed by a broadband noise masker. For each probe frequency, backward masking decreased approximately exponentially as the temporal interval between the probe and masker was increased. The rate of this exponential decay increased with increasing probe frequency. This latter result is consistent with Duifhuis' [J. Acoust. Soc. Am. 54, 1471-1488 (1973)] theory which attributes backward masking to the temporal overlap of cochlear responses to the probe and masker.

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.

The effect of broadband noise on the human brainstem auditory evoked response. I. Rate and intensity effects.

A series of experiments investigated the effects of continuous broadband noise (ipsilateral) on wave V of the click-evoked brainstem auditory evoked response (BAER). In general, a broadband noise masker increases the latency and decreases the amplitude of wave V. Varying both click and noise intensity, it was found that noise levels above about 40 dB SPL increase the latency and decrease the amplitude of wave V, regardless of click intensity. The effects of noise on wave V amplitude appear constant across click intensity, whereas the effects of a constant noise level on wave V latency decrease at higher click intensities. Both masking and adaptation increase wave V latency, but their combined effects are occlusive: rate-induced wave V latency shift decreases in the presence of continuous broadband noise. The clinical and theoretical implications of these findings are discussed.