Binaural processing model based on contralateral inhibition. II. Dependence on spectral parameters

Characterization of the decline in auditory nerve phase locking at high frequencies

The frequency dependence of phase locking in the auditory nerve influences various auditory coding mechanisms. The decline of phase locking with increasing frequency is commonly described by a low-pass filter. This study compares fitted low-pass filter parameters with the actual rate of phase locking decline. The decline is similar across studies and only 40 dB per decade, corresponding to the asymptotic decline of a second order filter.

Inferring the basis of binaural detection with a modified autoencoder

The binaural system utilizes interaural timing cues to improve the detection of auditory signals presented in noise. In humans, the binaural mechanisms underlying this phenomenon cannot be directly measured and hence remain contentious. As an alternative, we trained modified autoencoder networks to mimic human-like behavior in a binaural detection task. The autoencoder architecture emphasizes interpretability and, hence, we "opened it up" to see if it could infer latent mechanisms underlying binaural detection. We found that the optimal networks automatically developed artificial neurons with sensitivity to timing cues and with dynamics consistent with a cross-correlation mechanism. These computations were similar to neural dynamics reported in animal models. That these computations emerged to account for human hearing attests to their generality as a solution for binaural signal detection. This study examines the utility of explanatory-driven neural network models and how they may be used to infer mechanisms of audition.

Frequency dependence of sensitivity to interaural phase differences in pure tones

It is well established that in normal-hearing humans, the threshold of interaural time differences for pure tones increases dramatically above about 1300 Hz, only to become unmeasurable above 1400 Hz. However, physiological data and auditory models suggest that the actual decline in sensitivity is more gradual and only appears to be abrupt because the maximum of the psychometric function dips below the threshold proportion correct, e.g., 0.794. Published data only report thresholds at certain proportions correct but not the decline of proportions correct or of the sensitivity index d' with increasing frequencies. Here, we present pure-tone behavioral data obtained with a constant stimulus procedure. Seven of nine subjects showed proportions correct above 0.9 at 1300 Hz and virtually no sensitivity at 1500 Hz (proportion correct within 0.07 of chance level). This corresponds to a sensitivity decline of 46-78 dB/oct, much steeper than predicted by existing models or by the decline of phase locking of the auditory nerve fibers in animal data.

Perceptual similarity between piano notes: Simulations with a template-based perception model

In this paper, the auditory model developed by Dau, Kollmeier, and Kohlrausch [(1997). J. Acoust. Soc. Am. 102, 2892-2905] was used to simulate the perceptual similarity between complex sounds. As complex sounds, a set of piano recordings was used, whose perceptual similarity has recently been measured by Osses, Kohlrausch, and Chaigne [(2019). J. Acoust. Soc. Am. 146, 1024-1035] using a three-alternative forced-choice discrimination task in noise. To simulate this discrimination task, the auditory model required a new back-end stage, the central processor, which is preceded by several processing stages that are to a greater or lesser extent inspired by physiological aspects of the normal-hearing system. Therefore, a comprehensive review of the model parameters as used in the literature is given, indicating the fixed set of parameter values that is used in all simulations. Due to the perceptual relevance of the piano note onsets, this review includes an in-depth description of the auditory adaptation stage, the adaptation loops. A moderate to high correlation was found between the simulation results and existing experimental data.

Binaural frequency selectivity in humans

Several behavioural studies in humans have shown that listening to sounds with two ears that is binaural hearing, provides the human auditory system with extra information on the sound source that is not available when sounds are only perceived through one ear that is monaurally. Binaural processing involves the analysis of phase and level differences between the two ear signals. As monaural cochlea processing (in each ear) precedes the neural stages responsible for binaural processing properties it is reasonable to assume that properties of the cochlea may also be observed in binaural processing. A main characteristic of cochlea processing is its frequency selectivity. In psychoacoustics, there is an ongoing discussion on the frequency selectivity of the binaural auditory system. While some psychoacoustic experiments seem to indicate poorer frequency selectivity of the binaural system than that of the monaural processing others seem to indicate the same frequency selectivity for monaural and binaural processing. This study provides an overview of these seemingly controversial results and the different explanations that were provided to account for the different results.

Across-channel interaural-level-difference processing demonstrates frequency dependence

Accurate localization of complex sounds involves combining interaural information across frequencies to produce a single location percept. Interaural level differences (ILDs) are highly frequency dependent and it is unclear how the auditory system combines differing ILDs across frequency. Therefore, ILD just noticeable differences (JNDs) and intracranial lateralization were measured in young normal-hearing listeners using single- and multi-band stimuli. The bands were 300-ms, 10-Hz narrowband noises; the multi-band condition had three bands; they started and ended synchronously; they were located around three different frequency regions (750, 2000, or 4000 Hz); they had five different frequency separations that ranged from unresolved to resolved; the bands were dichotic with the same non-zero ILD (targets) or were diotic with zero ILD (interferers). Results showed single-band ILD JNDs were marginally frequency dependent. If unresolved diotic interferers were added, ILD JNDs increased greatly because of interaural decorrelation. If well-resolved diotic interferers were added, ILD JNDs were frequency dependent and the worst performance occurred when targets were near 1000 or 4000 Hz. This frequency dependence might be partially explained by ILD vs azimuth non-monotonicities for free-field sound sources in this frequency region. These results suggest that binaural processing models need revision for the processing of complex sounds.

A framework for testing and comparing binaural models

Auditory research has a rich history of combining experimental evidence with computational simulations of auditory processing in order to deepen our theoretical understanding of how sound is processed in the ears and in the brain. Despite significant progress in the amount of detail and breadth covered by auditory models, for many components of the auditory pathway there are still different model approaches that are often not equivalent but rather in conflict with each other. Similarly, some experimental studies yield conflicting results which has led to controversies. This can be best resolved by a systematic comparison of multiple experimental data sets and model approaches. Binaural processing is a prominent example of how the development of quantitative theories can advance our understanding of the phenomena, but there remain several unresolved questions for which competing model approaches exist. This article discusses a number of current unresolved or disputed issues in binaural modelling, as well as some of the significant challenges in comparing binaural models with each other and with the experimental data. We introduce an auditory model framework, which we believe can become a useful infrastructure for resolving some of the current controversies. It operates models over the same paradigms that are used experimentally. The core of the proposed framework is an interface that connects three components irrespective of their underlying programming language: The experiment software, an auditory pathway model, and task-dependent decision stages called artificial observers that provide the same output format as the test subject.

The effect of noise fluctuation and spectral bandwidth on gap detection

Experiment 1 investigated gap detection for random and low-fluctuation noise (LFN) markers as a function of bandwidth (25-1600 Hz), level [40 or 75 dB sound pressure level (SPL)], and center frequency (500-4000 Hz). Gap thresholds for random noise improved as bandwidth increased from 25 to 1600 Hz, but there were only minor effects related to center frequency and level. For narrow bandwidths, thresholds were lower for LFN than random markers; this difference extended to higher bandwidths at the higher center frequencies and was particularly large at high stimulus level. Effects of frequency and level were broadly consistent with the idea that peripheral filtering can increase fluctuation in the encoded LFN stimulus. Experiment 2 tested gap detection for 200-Hz-wide noise bands centered on 2000 Hz, using high-pass maskers to examine spread of excitation effects. Such effects were absent or minor for random noise markers and the 40-dB-SPL LFN markers. In contrast, some high-pass maskers substantially worsened performance for the 75-dB-SPL LFN markers. These results were consistent with an interpretation that relatively acute gap detection for the high-level LFN gap markers resulted from spread of excitation to higher-frequency auditory filters where the magnitude and phase characteristics of the LFN stimuli are better preserved.

The precedence effect in sound localization

In ordinary listening environments, acoustic signals reaching the ears directly from real sound sources are followed after a few milliseconds by early reflections arriving from nearby surfaces. Early reflections are spectrotemporally similar to their source signals but commonly carry spatial acoustic cues unrelated to the source location. Humans and many other animals, including nonmammalian and even invertebrate animals, are nonetheless able to effectively localize sound sources in such environments, even in the absence of disambiguating visual cues. Robust source localization despite concurrent or nearly concurrent spurious spatial acoustic information is commonly attributed to an assortment of perceptual phenomena collectively termed "the precedence effect," characterizing the perceptual dominance of spatial information carried by the first-arriving signal. Here, we highlight recent progress and changes in the understanding of the precedence effect and related phenomena.

Comparison of across-frequency integration strategies in a binaural detection model

Breebaart et al. [J. Acoust. Soc. Am. 110, 1089-1104 (2001)] reported that the masker bandwidth dependence of detection thresholds for an out-of-phase signal and an in-phase noise masker (N0Sπ) can be explained by principles of integration of information across critical bands. In this paper, different methods for such across-frequency integration process are evaluated as a function of the bandwidth and notch width of the masker. The results indicate that an "optimal detector" model assuming independent internal noise in each critical band provides a better fit to experimental data than a best filter or a simple across-frequency integrator model. Furthermore, the exponent used to model peripheral compression influences the accuracy of predictions in notched conditions.

The effects of noise-bandwidth, noise-fringe duration, and temporal signal location on the binaural masking-level difference

The effects of forward and backward noise fringes on binaural signal detectability were investigated. Masked thresholds for a 12-ms, 250-Hz, sinusoidal signal masked by Gaussian noise, centered at 250 Hz, with bandwidths from 3 to 201 Hz, were obtained in N(0)S(0) and N(0)S(π) configurations. The signal was (a) temporally centered in a 12-ms noise burst (no fringe), (b) presented at the start of a 600-ms noise burst (backward fringe), or (c) temporally centered in a 600-ms noise burst (forward-plus-backward fringe). For noise bandwidths between 3 and 75 Hz, detection in N(0)S(0) improved with the addition of a backward fringe, improving further with an additional forward fringe; there was little improvement in N(0)S(π). The binaural masking-level difference (BMLD) increased from 0 to 8 dB with a forward-plus-backward fringe as noise bandwidths increased to 100 Hz, increasing slightly to 10 dB at 201 Hz. This two-stage increase was less pronounced with a backward fringe. With no fringe, the BMLD was about 10-14 dB at all bandwidths. Performance appears to result from the interaction of across-time and across-frequency listening strategies and the possible effects of gain reduction and suppression, which combine in complex ways. Current binaural models are, as yet, unable to account fully for these effects.

The influence of different segments of the ongoing envelope on sensitivity to interaural time delays

The auditory system is sensitive to interaural timing disparities in the fine structure and the envelope of sounds, each contributing important cues for lateralization. In this study, psychophysical measurements were conducted with customized envelope waveforms in order to investigate the isolated effect of different segments of a periodic, ongoing envelope on lateralization. One envelope cycle was composed of the four segments attack flank, hold duration, decay flank, and pause duration, which were independently varied to customize the envelope waveform. The envelope waveforms were applied to a 4-kHz sinusoidal carrier, and just noticeable envelope interaural time differences were measured in six normal hearing subjects. The results indicate that attack durations and pause durations prior to the attack are the most important stimulus characteristics for processing envelope timing disparities. The results were compared to predictions of three binaural lateralization models based on the normalized cross correlation coefficient. Two of the models included an additional stage to mimic neural adaptation prior to binaural interaction, involving either a single short time constant (5 ms) or a combination of five time constants up to 500 ms. It was shown that the model with the single short time constant accounted best for the data.

Exploring the additivity of binaural and monaural masking release

Experiment 1 examined comodulation masking release (CMR) for a 700-Hz tonal signal under conditions of N(o)S(o) (noise and signal interaurally in phase) and N(o)S(π) (noise in phase, signal out of phase) stimulation. The baseline stimulus for CMR was either a single 24-Hz wide narrowband noise centered on the signal frequency [on-signal band (OSB)] or the OSB plus, a set of flanking noise bands having random envelopes. Masking noise was either gated or continuous. The CMR, defined with respect to either the OSB or the random noise baseline, was smaller for N(o)S(π) than N(o)S(o) stimulation, particularly when the masker was continuous. Experiment 2 examined whether the same pattern of results would be obtained for a 2000-Hz signal frequency; the number of flanking bands was also manipulated (two versus eight). Results again showed smaller CMR for N(o)S(π) than N(o)S(o) stimulation for both continuous and gated masking noise. The CMR was larger with eight than with two flanking bands, and this difference was greater for N(o)S(o) than N(o)S(π). The results of this study are compatible with serial mechanisms of binaural and monaural masking release, but they indicate that the combined masking release (binaural masking-level difference and CMR) falls short of being additive.

Stochastic model explains the role of excitation and inhibition in binaural sound localization in mammals

Interaural time differences (ITDs), the differences of arrival time of the sound at the two ears, provide a major cue for low-frequency sound localization in the horizontal plane. The first nucleus involved in the computation of ITDs is the medial superior olive (MSO). We have modeled the neural circuit of the MSO using a stochastic description of spike timing. The inputs to the circuit are stochastic spike trains with a spike timing distribution described by a given probability density function (beta density). The outputs of the circuit reproduce the empirical firing rates found in experiment in response to the varying ITD. The outputs of the computational model are calculated numerically and these numerical simulations are also supported by analytical calculations. We formulate a simple hypothesis concerning how sound localization works in mammals. According to this hypothesis, there is no array of delay lines as in the Jeffress' model, but the inhibitory input is shifted in time as a whole. This is consistent with experimental observations in mammals.

Effects of non-simultaneous masking on the binaural masking level difference

The present study sought to clarify the role of non-simultaneous masking in the binaural masking level difference for maskers that fluctuate in level. In the first experiment the signal was a brief 500-Hz tone, and the masker was a bandpass noise (100-2000 Hz), with the initial and final 200-ms bursts presented at 40-dB spectrum level and the inter-burst gap presented at 20-dB spectrum level. Temporal windows were fitted to thresholds measured for a range of gap durations and signal positions within the gap. In the second experiment, individual differences in out of phase (NoSπ) thresholds were compared for a brief signal in a gapped bandpass masker, a brief signal in a steady bandpass masker, and a long signal in a narrowband (50-Hz-wide) noise masker. The third experiment measured brief tone detection thresholds in forward, simultaneous, and backward masking conditions for a 50- and for a 1900-Hz-wide noise masker centered on the 500-Hz signal frequency. Results are consistent with comparable temporal resolution in the in phase (NoSo) and NoSπ conditions and no effect of temporal resolution on individual observers' ability to utilize binaural cues in narrowband noise. The large masking release observed for a narrowband noise masker may be due to binaural masking release from non-simultaneous, informational masking.

The role of across-frequency processes in dichotic listening conditions

In the bandwidening experiment with a diotic noise masker, an apparently wider critical bandwidth has often been reported when a dichotic signal (Spi) is used instead of a diotic signal (So). Two competing across-channel processes were proposed to account for this apparently wider critical bandwidth: (i) A detrimental across-channel effect reducing the binaural masking-level difference (BMLD) for broadband maskers and (ii) a beneficial integration of information across channels for narrowband maskers. The two hypotheses result in different predictions of the BMLD in the notched-noise experiment: According to the first hypothesis, the change in BMLD with notch width is determined by the level-dependence of the BMLD for a narrowband masker centered at the signal frequency, whereas the second hypothesis predicts that it is determined by the level-dependence of the BMLD for a broadband masker. To test the hypotheses, masked thresholds of a diotic or dichotic 500-Hz signal were measured for a diotic notched-noise masker as a function of notch width. In addition, thresholds were measured for a diotic broadband and narrowband masker as a function of masker level. The data indicate that neither of the two hypotheses is able to predict the continuous decrease in the BMLD as the notch width increases.

An evaluation of models for diotic and dichotic detection in reproducible noises

Several psychophysical models for masked detection were evaluated using reproducible noises. The data were hit and false-alarm rates from three psychophysical studies of detection of 500-Hz tones in reproducible noise under diotic (N0S0) and dichotic (N0Spi) conditions with four stimulus bandwidths (50, 100, 115, and 2900 Hz). Diotic data were best predicted by an energy-based multiple-detector model that linearly combined stimulus energies at the outputs of several critical-band filters. The tone-plus-noise trials in the dichotic data were best predicted by models that linearly combined either the average values or the standard deviations of interaural time and level differences; however, these models offered no predictions for noise-alone responses. The decision variables of more complicated temporal models, including the models of Dau et al. [(1996a). J. Acoust. Soc. Am. 99, 3615-3622] and Breebaart et al. [(2001a). J. Acoust. Soc. Am. 110, 1074-1088], were weakly correlated with subjects' responses. Comparisons of the dependencies of each model on envelope and fine-structure cues to those in the data suggested that dependence upon both envelope and fine structure, as well as an interaction between them, is required to predict the detection results.

Lateralization of stimuli with independent fine-structure and envelope-based temporal disparities

Psychoacoustic experiments were conducted to investigate the role and interaction of fine-structure and envelope-based interaural temporal disparities. A computational model for the lateralization of binaural stimuli, motivated by recent physiological findings, is suggested and evaluated against the psychoacoustic data. The model is based on the independent extraction of the interaural phase difference (IPD) from the stimulus fine-structure and envelope. Sinusoidally amplitude-modulated 1-kHz tones were used in the experiments. The lateralization from either carrier (fine-structure) or modulator (envelope) IPD was matched with an interaural level difference, revealing a nearly linear dependence for both IPD types up to 135 degrees , independent of the modulation frequency. However, if a carrier IPD was traded with an opposed modulator IPD to produce a centered sound image, a carrier IPD of 45 degrees required the largest opposed modulator IPD. The data could be modeled assuming a population of binaural neurons with a physiological distribution of the best IPDs clustered around 45 degrees -50 degrees . The model was also used to predict the perceived lateralization of previously published data. Subject-dependent differences in the perceptual salience of fine-structure and envelope cues, also reported previously, could be modeled by individual weighting coefficients for the two cues.

Interaural time discrimination of envelopes carried on high-frequency tones as a function of level and interaural carrier mismatch

OBJECTIVES

The present study investigated interaural time discrimination for binaurally mismatched carrier frequencies in listeners with normal hearing. One goal of the investigation was to gain insights into binaural hearing in patients with bilateral cochlear implants, where the coding of interaural time differences (ITDs) may be limited by mismatches in the neural populations receiving stimulation on each side.

DESIGN

Temporal envelopes were manipulated to present low frequency timing cues to high-frequency auditory channels. Carrier frequencies near 4 kHz were amplitude modulated at 128 Hz via multiplication with a half-wave rectified sinusoid, and that modulation was either in-phase across ears or delayed to one ear. Detection thresholds for nonzero ITDs were measured for a range of stimulus levels and a range of carrier frequency mismatches. Data were also collected under conditions designed to limit cues based on stimulus spectral spread, including masking and truncation of sidebands associated with modulation.

RESULTS

Listeners with normal hearing can detect ITDs in the face of substantial mismatches in carrier frequency across ears.

CONCLUSIONS

The processing of ITDs in listeners with normal hearing is likely based on spread of excitation into binaurally matched auditory channels. Sensitivity to ITDs in listeners with cochlear implants may depend on spread of current that results in the stimulation of neural populations that share common tonotopic space bilaterally.

A computational model of human auditory signal processing and perception

A model of computational auditory signal-processing and perception that accounts for various aspects of simultaneous and nonsimultaneous masking in human listeners is presented. The model is based on the modulation filterbank model described by Dau et al. [J. Acoust. Soc. Am. 102, 2892 (1997)] but includes major changes at the peripheral and more central stages of processing. The model contains outer- and middle-ear transformations, a nonlinear basilar-membrane processing stage, a hair-cell transduction stage, a squaring expansion, an adaptation stage, a 150-Hz lowpass modulation filter, a bandpass modulation filterbank, a constant-variance internal noise, and an optimal detector stage. The model was evaluated in experimental conditions that reflect, to a different degree, effects of compression as well as spectral and temporal resolution in auditory processing. The experiments include intensity discrimination with pure tones and broadband noise, tone-in-noise detection, spectral masking with narrow-band signals and maskers, forward masking with tone signals and tone or noise maskers, and amplitude-modulation detection with narrow- and wideband noise carriers. The model can account for most of the key properties of the data and is more powerful than the original model. The model might be useful as a front end in technical applications.

Binaural processing of modulated interaural level differences

Two experiments are presented that measure the acuity of binaural processing of modulated interaural level differences (ILDs) using psychoacoustic methods. In both experiments, dynamic ILDs were created by imposing an interaurally antiphasic sinusoidal amplitude modulation (AM) signal on high-frequency carriers, which were presented over headphones. In the first experiment, the sensitivity to dynamic ILDs was measured as a function of the modulation frequency using puretone, and interaurally correlated and uncorrelated narrow-band noise carriers. The intrinsic interaural level fluctuations of the uncorrelated noise carriers raised the ILD modulation detection thresholds with respect to the pure-tone carriers. The diotic fluctuations of the correlated noise carriers also caused a small increase in the thresholds over the pure-tone carriers, particularly with low ILD modulation frequencies. The second experiment investigated the modulation frequency selectivity in dynamic ILD processing by imposing an interaurally uncorrelated bandpass noise AM masker in series with the interaurally antiphasic AM signal on a pure-tone carrier. By varying the masker center frequencies relative to the signal modulation frequency, broadly tuned, bandpass-shaped patterns were obtained. Simulations with an existing binaural model show that a low-pass filter to limit the binaural temporal resolution is not sufficient to predict the results of the experiments.

The effect of stimulus intensity on the right ear advantage in dichotic listening

The dichotic listening test is non-invasive behavioural technique to study brain lateralization and it has been shown, that its results can be systematically modulated by varying stimulation properties (bottom-up effects) or attentional instructions (top-down effects) of the testing procedure. The goal of the present study was to further investigate the bottom-up modulation, by examining the effect of differences in the right or left ear stimulus intensity on the ear advantage. For this purpose, interaural intensity difference were gradually varied in steps of 3 dB from -21 dB in favour of the left ear to +21 dB in favour of the right ear, also including a no difference baseline condition. Thirty-three right-handed adult participants with normal hearing acuity were tested. The dichotic listening paradigm was based on consonant-vowel stimuli pairs. Only pairs with the same voicing (voice or non-voiced) of the consonant sound were used. The results showed: (a) a significant right ear advantage (REA) for interaural intensity differences from 21 to -3 dB, (b) no ear advantage (NEA) for the -6 dB difference, and (c) a significant left ear advantage (LEA) for differences form -9 to -21 dB. It is concluded that the right ear advantage in dichotic listening to CV syllables withstands an interaural intensity difference of -9 dB before yielding to a significant left ear advantage. This finding could have implications for theories of auditory laterality and hemispheric asymmetry for phonological processing.

Perception of across-frequency interaural level differences

The interaural level difference (ILD) is an important cue for the localization of sound sources. Just noticeable differences (JND) in ILD were measured in 12 normal hearing subjects for uncorrelated noise bands with a bandwidth of 13 octave and a different center frequency in both ears. In one ear the center frequency was either 250, 500, 1000, or 4000 Hz. In the other ear, a frequency shift of 0, 16, 13, or 1 octave was introduced. JNDs in ILD for unshifted, uncorrelated noise bands of 13 octave width were 2.6, 2.6, 2.5, and 1.4 dB for 250, 500, 1000, and 4000 Hz, respectively. Averaged over all shifts, JNDs decreased significantly with increasing frequency. For the shifted conditions, JNDs increased significantly with increasing shift. Performance on average worsened by 0.5, 0.9, and 1.5 dB for shifts of 16, 13, and 1 octave. Though performance decreases, the just noticeable ILDs for the shifted conditions were still in a range usable for lateralization. This has implications for signal processing algorithms for bilateral bimodal hearing instruments and the fitting of bilateral cochlear implants.

Evidence specifically favoring the equalization-cancellation theory of binaural unmasking

Three experiments investigated the roles of interaural correlation (rho) and of the monaural power spectrum in the detection and discrimination of narrow-band-noise signals (462-539 Hz) in broadband maskers (0-3 kHz). The power and rho of the target band were independently controlled, while the flanking noise was fixed and diotic. Experiments 1 and 2 involved rho and power values that would be produced by specific values of signal-to-noise ratio (SNR) in the NoSpi binaural configuration. Listeners were required to discriminate different SNRs via a 2I-FC loudness-discrimination task. At low reference SNRs, changes in rho fully accounted for listeners' performance, but as reference SNR increased, additional energy in the target band played an increasing role. Experiment 2 showed that at these higher SNRs the combination of information from the power spectrum and rho was superadditive and could not be explained by simple signal-detection models. The equalization-cancellation (EC) theory would explain these data using the output from interaural cancellation, Y, rather than rho. Experiment 3 attempted to foil binaural processing, by fixing either rho or Y across intervals. Consistent with EC theory, when Y was fixed, the contribution of the binaural system appeared negligible, while fixing rho did not have this effect.

Interaural fluctuations and the detection of interaural incoherence. III. Narrowband experiments and binaural models

In the first two articles of this series, reproducible noises with a fixed value of interaural coherence (0.992) were used to study the human ability to detect interaural incoherence. It was found that incoherence detection is strongly correlated with fluctuations in interaural differences, especially for narrow noise bandwidths, but it remained unclear what function of the fluctuations best agrees with detection data. In the present article, ten different binaural models were tested against detection data for 14- and 108-Hz bandwidths. These models included different types of binaural processing: independent-interaural-phase-difference/interaural-level-difference, lateral-position, and short-term cross-correlation. Several preprocessing transformations of the interaural differences were incorporated: compression of binaural cues, temporal averaging, and envelope weighting. For the 14-Hz bandwidth data, the most successful model postulated that incoherence is detected via fluctuations of interaural phase and interaural level processed by independent centers. That model correlated with detectability at r=0.87. That model proved to be more successful than short-term cross-correlation models incorporating standard physiologically-based model features (r=0.78). For the 108-Hz bandwidth data, detection performance varied much less among different waveforms, and the data were less able to distinguish between models.

Binaural comodulation masking release: effects of masker interaural correlation

Binaural detection was examined for a signal presented in a narrow band of noise centered on the on-signal masking band (OSB) or in the presence of flanking noise bands that were random or comodulated with respect to the OSB. The noise had an interaural correlation of 1.0 (No), 0.99 or 0.95. In No noise, random flanking bands worsened Spi detection and comodulated bands improved Spi detection for some listeners but had no effect for other listeners. For the 0.99 or 0.95 interaural correlation conditions, random flanking bands were less detrimental to Spi detection and comodulated flanking bands improved Spi detection for all listeners. Analyses based on signal detection theory indicated that the improvement in Spi thresholds obtained with comodulated bands was not compatible with an optimal combination of monaural and binaural cues or to across-frequency analyses of dynamic interaural phase differences. Two accounts consistent with the improvement in Spi thresholds in comodulated noise were (1) envelope information carried by the flanking bands improves the weighting of binaural cues associated with the signal; (2) the auditory system is sensitive to across-frequency differences in ongoing interaural correlation.

Influence of dynamic compression on directional hearing in the horizontal plane

OBJECTIVE

The purpose of this study was to assess the influence of dynamic compression on directional hearing in the frontal horizontal plane.

DESIGN

Compression schemes with various compression ratios and attack times were created by using a digital signal processor. The influence of compression on isolated interaural level differences (ILDs) and interaural time differences (ITDs) was examined in discrimination experiments. The combination of ILDs and ITDs was investigated with measurements of localization based on head-related transfer functions. Five normal listeners and seven hearing-impaired subjects participated in the study.

RESULTS

Dynamic compression revealed substantial effects on discrimination of ILDs but not on ITDs. With higher compression ratios and shorter attack times, just-noticeable differences for ILDs increased. Lateralization via head-related transfer functions was predominantly affected by compression in the higher frequency range because for lower frequencies, interaural time cues dominate over interaural level cues. The effect of compression was similar for normal and hearing-impaired listeners, but the latter performed worse on almost all tasks, especially with ITD discrimination at 4000 Hz.

CONCLUSIONS

The influence of compression ratio and attack time could be shown for ILDs, but the effect was rather small, at least for the low compression ratios commonly used in hearing aids. The influence further decreased when attack time was prolonged. The dramatic impairment of the discrimination of ITDs with the hearing-impaired subjects in the high-frequency range suggests that they primarily rely on ILD cues.

Prediction of speech intelligibility in spatial noise and reverberation for normal-hearing and hearing-impaired listeners

Binaural speech intelligibility of individual listeners under realistic conditions was predicted using a model consisting of a gammatone filter bank, an independent equalization-cancellation (EC) process in each frequency band, a gammatone resynthesis, and the speech intelligibility index (SII). Hearing loss was simulated by adding uncorrelated masking noises (according to the pure-tone audiogram) to the ear channels. Speech intelligibility measurements were carried out with 8 normal-hearing and 15 hearing-impaired listeners, collecting speech reception threshold (SRT) data for three different room acoustic conditions (anechoic, office room, cafeteria hall) and eight directions of a single noise source (speech in front). Artificial EC processing errors derived from binaural masking level difference data using pure tones were incorporated into the model. Except for an adjustment of the SII-to-intelligibility mapping function, no model parameter was fitted to the SRT data of this study. The overall correlation coefficient between predicted and observed SRTs was 0.95. The dependence of the SRT of an individual listener on the noise direction and on room acoustics was predicted with a median correlation coefficient of 0.91. The effect of individual hearing impairment was predicted with a median correlation coefficient of 0.95. However, for mild hearing losses the release from masking was overestimated.

The effect of diotic and dichotic level-randomization on the binaural masking-level difference

Detection thresholds for tones in narrow-band noise were measured for two binaural configurations: N(o)S(o) and N(o)S(pi). The 30-Hz noise band had a mean overall level of 65 dB SPL and was centered on 250, 500, or 5000 Hz. Signals and noise were simultaneously gated for 500, 110, or 20 ms. Three conditions of level randomization were tested: (1) no randomization; (2) diotic randomization--the stimulus level (common to both ears) was randomly chosen from an uniformly distributed 40-dB range every presentation interval; and (3) dichotic randomization--the stimulus levels for each ear were each independently and randomly chosen from the 40-dB range. Regardless of binaural configuration, level randomization had small effects on thresholds at 500 and 110 ms, implying that binaural masking-level differences (BMLDs) do not depend on interaural level differences for individual stimuli. For 20-ms stimuli, both diotic and dichotic randomization led to markedly poorer performance than at 500- and 110-ms durations; BMLDs diminished with no randomization and dichotic randomization but not with diotic randomization. The loss of BMLDs at 20 ms, with degrees-of-freedom (2WT) approximately 1, implies that changes in intracranial parameters occurring during the course of the observation interval are necessary for BMLDs when mean-level and mean-intracranial-position cues have been made unhelpful.

Monaural and interaural intensity discrimination: level effects and the "binaural advantage"

This study examined whether the level effects seen in monaural intensity discrimination (Weber's law and the "near miss") in a two-interval task are also observed in discrimination of interaural intensity differences (IIDs) in a single-interval task. Both tasks were performed for various standard levels of 4-kHz pure tones and broadband noise. The Weber functions (10 log deltaI/I versus I in dB) in the monaural and binaural conditions were parallel. For noise, the Weber functions had slopes close to zero (Weber's law) while the Weber functions for the tones had a mean slope of -0.089 (near miss). The near miss for the monaural and binaural tasks with tones was eliminated when a high-pass masker was gated with the listening intervals. The near-miss was also observed for 250- and 1000-Hz tones in the binaural task despite overall decreased sensitivity to changes in IID at 1000 Hz. The binaural thresholds showed a small (about 2-dB) advantage over monaural thresholds only in the broadband noise conditions. More important, however, is the fact that the level effects seen monaurally are also seen binaurally. This suggests that the basic mechanisms responsible for Weber's law and the near miss are common to monaural and binaural processing.

The across frequency independence of equalization of interaural time delay in the equalization-cancellation model of binaural unmasking

The equalization stage in the equalization-cancellation model of binaural unmasking compensates for the interaural time delay (ITD) of a masking noise by introducing an opposite, internal delay [N. I. Durlach, in Foundations of Modern Auditory Theory, Vol. II., edited by J. V. Tobias (Academic, New York, 1972)]. Culling and Summerfield [J. Acoust. Soc. Am. 98, 785-797 (1995)] developed a multi-channel version of this model in which equalization was "free" to use the optimal delay in each channel. Two experiments were conducted to test if equalization was indeed free or if it was "restricted" to the same delay in all channels. One experiment measured binaural detection thresholds, using an adaptive procedure, for 1-, 5-, or 17-component tones against a broadband masking noise, in three binaural configurations (N0S180, N180S0, and N90S270). The thresholds for the 1-component stimuli were used to normalize the levels of each of the 5- and 17-component stimuli so that they were equally detectable. If equalization was restricted, then, for the 5- and 17-component stimuli, the N90S270 and N180S0 configurations would yield a greater threshold than the N0S180 configurations. No such difference was found. A subsequent experiment measured binaural detection thresholds, via psychometric functions, for a 2-component complex tone in the same three binaural configurations. Again, no differential effect of configuration was observed. An analytic model of the detection of a complex tone showed that the results were more consistent with free equalization than restricted equalization, although the size of the differences was found to depend on the shape of the psychometric function for detection.

The benefit of binaural hearing in a cocktail party: effect of location and type of interferer

The "cocktail party problem" was studied using virtual stimuli whose spatial locations were generated using anechoic head-related impulse responses from the AUDIS database [Blauert et al., J. Acoust. Soc. Am. 103, 3082 (1998)]. Speech reception thresholds (SRTs) were measured for Harvard IEEE sentences presented from the front in the presence of one, two, or three interfering sources. Four types of interferer were used: (1) other sentences spoken by the same talker, (2) time-reversed sentences of the same talker, (3) speech-spectrum shaped noise, and (4) speech-spectrum shaped noise, modulated by the temporal envelope of the sentences. Each interferer was matched to the spectrum of the target talker. Interferers were placed in several spatial configurations, either coincident with or separated from the target. Binaural advantage was derived by subtracting SRTs from listening with the "better monaural ear" from those for binaural listening. For a single interferer, there was a binaural advantage of 2-4 dB for all interferer types. For two or three interferers, the advantage was 2-4 dB for noise and speech-modulated noise, and 6-7 dB for speech and time-reversed speech. These data suggest that the benefit of binaural hearing for speech intelligibility is especially pronounced when there are multiple voiced interferers at different locations from the target, regardless of spatial configuration; measurements with fewer or with other types of interferers can underestimate this benefit.

Transformations in processing interaural time differences between the superior olivary complex and inferior colliculus: beyond the Jeffress model

Interaural time differences (ITDs) are used to localize sounds and improve signal detection in noise. Encoding ITDs in neurons depends on specialized mechanisms for comparing inputs from the two ears. Most studies have emphasized how the responses of ITD-sensitive neurons are consistent with the tenets of the Jeffress model. The Jeffress model uses neuronal coincidence detectors that compare inputs from both sides and delay lines so that different neurons achieve coincidence at different ITDs. Although Jeffress-type models are successful at predicting sensitivity to ITDs in humans, in many respects they are a limited representation of the responses seen in neurons. In the superior olivary complex (SOC), ITD-sensitive neurons are distributed across both the medial (MSO) and lateral (LSO) superior olives. Similar response types are found in neurons sensitive to ITDs in two signal types: low-frequency sounds and envelopes of high-frequency sounds. Excitatory-excitatory interactions in the MSO are associated with peak-type responses, and excitatory-inhibitory interactions in the LSO are associated with trough-type responses. There are also neurons with responses intermediate between peak- and trough-type. In the inferior colliculus (IC), the same basic types remain, presumably due to inputs arising from the MSO and LSO. Using recordings from the SOC and IC, we describe how the response types can be described within a continuum that extends to very large values of ITD, and compare the functional organization at the two levels.

Binaural processing model based on contralateral inhibition. III. Dependence on temporal parameters

This paper and two accompanying papers [Breebaart et al., J. Acoust. Soc. Am. 110, 1074-1088 (2001); 110, 1089-1104 (2001)] describe a computational model for the signal processing of the binaural auditory system. The model consists of several stages of monaural and binaural preprocessing combined with an optimal detector. Simulations of binaural masking experiments were performed as a function of temporal stimulus parameters and compared to psychophysical data adapted from literature. For this purpose, the model was used as an artificial observer in a three-interval, forced-choice procedure. All model parameters were kept constant for all simulations. Model predictions were obtained as a function of the interaural correlation of a masking noise and as a function of both masker and signal duration. Furthermore, maskers with a time-varying interaural correlation were used. Predictions were also obtained for stimuli with time-varying interaural time or intensity differences. Finally, binaural forward-masking conditions were simulated. The results show that the combination of a temporal integrator followed by an optimal detector in the time domain can account for all conditions that were tested, except for those using periodically varying interaural time differences (ITDs) and those measuring interaural correlation just-noticeable differences (jnd's) as a function of bandwidth.

Binaural processing model based on contralateral inhibition. I. Model structure

This article presents a quantitative binaural signal detection model which extends the monaural model described by Dau et al. [J. Acoust. Soc. Am. 99, 3615-3622 (1996)]. The model is divided into three stages. The first stage comprises peripheral preprocessing in the right and left monaural channels. The second stage is a binaural processor which produces a time-dependent internal representation of the binaurally presented stimuli. This stage is based on the Jeffress delay line extended with tapped attenuator lines. Through this extension, the internal representation codes both interaural time and intensity differences. In contrast to most present-day models, which are based on excitatory-excitatory interaction, the binaural interaction in the present model is based on contralateral inhibition of ipsilateral signals. The last stage, a central processor, extracts a decision variable that can be used to detect the presence of a signal in a detection task, but could also derive information about the position and the compactness of a sound source. In two accompanying articles, the model predictions are compared with data obtained with human observers in a great variety of experimental conditions.