Model Of Binaural Interaction Research Articles

Functional rate-code models of the auditory brainstem for predicting lateralization and discrimination data of human binaural perception.

This paper presents a rate-code model of binaural interaction inspired by recent neurophysiological findings. The model consists of a peripheral part and a binaural part. The binaural part is composed of models of the medial superior olive (MSO) and the lateral superior olive (LSO), which are parts of the auditory brainstem. The MSO and LSO model outputs are preprocessed in the interaural time difference (ITD) and interaural level difference (ILD) central stages, respectively, which give absolute values of the predicted lateralization at their outputs, allowing a direct comparison with psychophysical data. The predictions obtained with the MSO and LSO models are compared with subjective data on the lateralization of pure tones and narrowband noises, discrimination of the ITD and ILD, and discrimination of the phase warp. The lateralization and discrimination experiments show good agreement with the subjective data. In the case of the phase-warp experiment, the models agree qualitatively with the subjective data. The results demonstrate that rate-code models of MSO and LSO can be used to explain psychophysical data considering lateralization and discrimination based on binaural cues.

Spectrotemporal window of binaural integration in auditory object formation

Binaural integration of interaural temporal information is essential for sound source localization and segregation. Current models of binaural interaction have shown that accurate sound localization in the horizontal plane depends on the resolution of phase ambiguous information by across-frequency integration. However, as such models are mostly static, it is not clear how proximate in time binaural events in different frequency channels should occur to form an auditory object with a unique lateral position. The present study examined the spectrotemporal window required for effective integration of binaural cues across frequency to form the perception of a stationary position. In Experiment 1, listeners judged whether dichotic frequency-modulated (FM) sweeps with a constant large nominal interaural delay (1500 μs), whose perceived laterality was ambiguous depending on the sweep rate (1500, 3000, 6000, and 12,000 Hz/s), produced a percept of continuous motion or a stationary image. Motion detection performance, indexed by d-prime (d') values, showed a clear effect of sweep rate, with auditory motion effects most pronounced for low sweep rates, and a punctate stationary image at high rates. Experiment 2 examined the effect of modulation rate (0.5, 3, 20, and 50 Hz) on lateralizing sinusoidally frequency-modulated (SFM) tones to confirm the effect of sweep rate on motion detection, independent of signal duration. Lateralization accuracy increased with increasing modulation rate up to 20 Hz and saturated at 50 Hz, with poorest performance occurring below 3 Hz depending on modulator phase. Using the transition point where percepts changed from motion to stationary images, we estimated a spectrotemporal integration window of approximately 150 ms per octave required for effective integration of interaural temporal cues across frequency channels. A Monte Carlo simulation based on a cross-correlation model of binaural interaction predicted 90% of the variance on perceptual motion detection performance as a function of FM sweep rate. Findings suggest that the rate of frequency channel convergence of binaural cues is essential to binaural lateralization.

Time and intensity with Tino

Over a period of decades Constantine (Tino) Trahiotis has been a contemplative and creative scholar, an excellent teacher, a helpful collaborator, as well as a loyal friend. Tino, along with Bob Bilger and Erv Hafter, provided this author's introduction to the broader science of hearing beyond the particular perspective of my graduate training. At the same time, my attempts to interpret models of binaural interaction with Tino led to the review chapters on binaural modeling that the two of us jointly authored in the 1990s. Tino's demonstration of the unexpected dependence on bandwidth of the laterality of bands of noise presented with interaural temporal differences of large magnitude motivated me to think about the tradeoff between “straightness” and “centrality,” which appears to be helpful in considering the lateralization of many binaural stimuli. Finally, Tino, along with Les Bernstein, insisted on a local implementation of my models in their laboratory which became one source (along with significant others) for the binaural modeling toolbox developed by Michael Akeroyd that is presently in wide circulation. This talk will review these facets of my life with Tino, and comment on my current perspectives on related issues. [Work supported by DARPA and Cisco Research.]

Lateralization based on interaural differences in the second-order amplitude modulator

Second-order amplitude modulation is a relatively slow variation of the modulation depth of a first-order amplitude modulation with higher frequency. In contrast to first-order modulation, which appears as a physical component in the stimulus spectrum after half-wave rectification, second-order modulation is not necessarily demodulated by the auditory periphery. For binaural processing of second-order amplitude modulated stimuli it is unknown whether interaural time differences (ITDs) in the second-order modulation result in a lateralized percept. Thus, second-order modulation can serve as a tool to investigate whether demodulation of interaurally delayed components is a prerequisite for lateralization. In most of the psychoacoustic experiments presented here, a 25 Hz sinusoidally amplitude-modulated (SAM) 160 Hz tone was either transposed to 4 kHz by half-wave rectifying this SAM waveform before multiplication with a 4 kHz tone (TSAM), or by adding an offset before multiplication (SAMAM). The experiments revealed an inability to lateralize the SAMAM based on ITDs in the 25 Hz component, whereas subjects could lateralize the TSAM. Given that only the TSAM results in a demodulated 25 Hz component after peripheral auditory processing, this result supports the hypothesis that demodulation is a prerequisite for lateralization, which has consequences for temporal modulation processing in models of binaural interaction.

Cross-Modulation Interference With Lateralization of Mixed-Modulated Waveforms

This study investigated the ability to use spatial information in mixed-modulated (MM) sounds containing concurrent frequency-modulated (FM) and amplitude-modulated (AM) sounds by exploring patterns of interference when different modulation types originated from different loci as may occur in a multisource acoustic field. Interaural delay thresholds were measured from 5 normal-hearing adults for an AM sound in the presence of interfering FM and vice versa as a function of interferer modulation rate. In addition, the effects of near versus remote interferer rates, and fixed versus randomized interferer interaural delay, were investigated. AM interfered with lateralization of FM at all modulation rates. However, the FM interfered with AM lateralization only when the FM rate was higher than the AM rate. This rate asymmetry was surprising given the prevalence of low-frequency dominance in lateralization, but was predicted by a cross-correlation model of binaural interaction. Effects were similar for fixed and randomized interferer interaural delays. The results suggest that in multisource environments, sources containing different modulation types significantly interfere with localization in complex ways that reveal interactions between modulation type and rate. These findings contribute to the understanding of auditory object formation and localization.

Computational Auditory Scene Analysis: Principles, Algorithms and Applications

First Page

A History of Binaural Interaction Modeling - Walks with Jens Blauert

Jens Blauert has been cut an impressively wide path through diverse areas of binaural and spatial hearing. He has consistently integrated viewpoints based on empirical results, intuitive interpretations, and mathematical and computational models of binaural interaction and spatial hearing. This talk focuses on the development of ideas about binaural interaction as represented in quantitative models and the strong influence that Blauert exerted on these models. These influences start with simple models of the extraction and use of interaural time and level differences, include mechanisms that incorporate complex interactions of these differences, and lead to models that are appropriate for the interpretation of complex acoustic scenes. Finally, some current modeling activities will be outlined and related to these historical developments. [Work supported by the US National Institutues of Health DC00100.]

Lateralization of linear and logarithmic frequency-modulated (FM) sweeps at high frequencies

The ability to lateralize linear and logarithmic frequency‐modulated (FM) sweeps at high frequencies was investigated using a 2IFC adaptive block design. The FM stimulus swept from 3 kHz to 8 kHz with the onset frequency roved by 15% on each presentation. The cue to be detected was an interaural delay in the waveform’s fine‐structure (i.e., delay in instantaneous frequency) that varied on each trial according to a 2‐down 1‐up adaptive rule. The waveforms to the left and right channels had simultaneous onset/offset envelopes. The stimulus duration (0.5, 1, 5, 10, 25, or 50 ms) and thus the sweep rate, as well as the sweep direction (up or down), were additional parameters of the study. Thresholds were higher than expected from studies of sinusoidal FM and ranged from approximately 100 μs for the 0.5 ms sweep to near chance for durations exceeding 10 ms. Sweep direction and type (log versus linear) did not substantially affect thresholds. Implications of these findings for models of binaural interaction at high frequencies and within‐channel mechanisms of FM‐to‐AM transduction will be discussed. [Work supported by NSF.]

Detection of large interaural delays and its implication for models of binaural interaction.

The interaural time difference (ITD) is a major cue to sound localization along the horizontal plane. The maximum natural ITD occurs when a sound source is positioned opposite to one ear. We examined the ability of owls and humans to detect large ITDs in sounds presented through headphones. Stimuli consisted of either broad or narrow bands of Gaussian noise, 100 ms in duration. Using headphones allowed presentation of ITDs that are greater than the maximum natural ITD. Owls were able to discriminate a sound leading to the left ear from one leading to the right ear, for ITDs that are 5 times the maximum natural delay. Neural recordings from optic-tectum neurons, however, show that best ITDs are usually well within the natural range and are never as large as ITDs that are behaviorally discriminable. A model of binaural crosscorrelation with short delay lines is shown to explain behavioral detection of large ITDs. The model uses curved trajectories of a cross-correlation pattern as the basis for detection. These trajectories represent side peaks of neural ITD-tuning curves and successfully predict localization reversals by both owls and human subjects.

An auditory illusion predicted from a weighted cross-correlation model of binaural interaction.

In humans, the lateral movement of an acoustic source produces dynamic changes in the relative sound-pressure level and time of arrival of the acoustic wave at the 2 ears. The dynamic nature of these cues is assumed to play an important role in the perception of lateral motion. A phenomenon of auditory motion is reported whose lateral direction and relative velocity may be specified while interaural differences are kept constant. The stimulus producing this percept is a narrowband wave-form whose instantaneous bandwidth is a cosine function of time. This phenomenon is predicted from a model of cross-correlation that estimates the running position of an image from a weighted combination of 2 variables: (a) magnitude of interaural delay, with smaller delays receiving more weight, and (b) consistency of interaural information across frequency.

Modeling of binaural discrimination of multiple sound sources: A contribution to the development of a “cocktail party processor”

The human hearing system when being stimulated simultaneously by narrow‐banded signals from two loudspeakers at different locations is able to determine the directions of both sound sources. However, conventional models of binaural interaction which are based on interaural cross correlation in critical bands fail in discriminating the source directions. The reason is that the cross‐correlation functions at the output of the model display severe fluctuations in time. A revised binaural cross‐correlation model that contains an additional “recomputation mechanism” will be presented. This mechanism estimates directions and energies of sound sources from cross‐correlation functions fluctuating in time. Using this mechanism the results of relevant auditory experiments can be reproduced. This model has been applied to construct a cocktail party processor, which is able to enhance the signal‐to‐noise ratio considerably.

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.

Lateral-position-based models of interaural discrimination.

This letter investigates the hypothesis that lateral position is the only cue available for interaural discrimination experiments using 500-Hz stimuli. The discussion of this hypothesis is in the context of comparisons of the experimental data to predictions of the "position-variable" model of binaural interaction. The model predicts the mean and variance of the subjective lateral position of stimuli used in the discrimination experiments, assuming that discrimination performance is based on optimal processing of this subjective position. To the extent that the laterality predictions of the model are accurate, data that are inconsistent with its predictions would also be problematical for any model based on the subjective laterality of a single binaural image. The predictions (at least qualitatively) describe much of the observed experimental data, including a number of results that have not been addressed by any previous theory. Nevertheless, the observed performance is significantly better than the corresponding predictions for three types of experiments in which the utility of the position cue has been eliminated by experimental design. We believe that our results indicate that changes in lateral position are the primary cue in most interaural discrimination experiments, but that secondary attributes of the perceptual images can be useful when performance based on position alone would be poor.

A general modification for models of binaural interaction

Although many studies have demonstrated the relationshio between lateralization and binaural detection, current models of binaural interaction are unable to predict performance in both tasks with the same set of values for the model parameters. By incorporating a mechanism which low‐pass filters (averages) interaural time and intensity differences into a simple, narrow‐band model of binaural interaction, results from interaural time and intensity discrimination tests can be used to predict correlation discrimination and NoSπ detection results. Evidence for such a low‐pass mechanism has most recently been reported by Grantham and Wightman [J. Acoust. Soc. Am. 72, 1178–1184 (1982)]. For both normal listeners and hearing‐impaired listeners with a variety of different losses, the predictions of the augmented model suggest that binaural hearing in correlation discrimination and NoSπ detection tasks can be characterized by a listener's sensitivity to interaural time and intensity differences. [Work supported by NIH Grant NS10916.]

Lateralization of complex sounds

At one time it was felt that interaural differences of time could affect the localization only of low frequencies. Psychoacoustic support for this view came from observations with sinusoidal stimuli that showed a total insensitivity to interaural delays in tones above about 1.5 kHz. In recent years, however, it has become increasingly clear that listeners can detect interaural delay in signals of high-frequency content, as long as the spectrum is complex. For example, performance can be excellent when the signal is a high-frequency carrier that has been modulated in amplitude. The purpose of this paper is to discuss functional differences between simple and complex sounds, with an eye toward development of a model of binaural interaction that accounts for the responses to both. Toward this end, special attention will be paid to stimulus factors that differentiate the two classes. In the preceding paper, Yost compares the responses to interaural differences of time in the stimulus onset and in the waveform. That notion will be expanded here in a model which attributes a spectral role to onsets and which visualizes both individual peaks in the envelopes of high frequencies and the oscillations of low frequencies as successions of mini-onsets.

A predictive model for binaural advantages in speech intelligibility

This paper presents a model of binaural advantages in speech intelligibility applicable to anechoic conditions in which speech is masked by a single noise source, with speech and noise sources both located in the horizontal plane. First, frequency- and azimuth-dependent sound-pressure transformations (from Shaw's summary of amplitude measurements and Woodworth's equation for phase) are applied to obtain third-octave signal and noise spectra at each ear. Next (following Levitt and Rabiner), the effective S/N ratio in each band is increased by the masking-level difference, which is approximated with an equation (from Colburn's model of binaural interaction) that describes MLD dependence on relevant stimulus parameters. Monaural and binaural performance are then predicted using Articulation Theory (Kryter). Predictions are in reasonable agreement with data from seventeen studies of the effects of source azimuth and listening mode (monaural or binaural) on intelligibility. The model provides a quantitative framework for analyzing binaural advantages and for assessing the practical benefits of binaural hearing. [Work supported by NIH.]

A model of binaural interaction

We present here a model which simulates some recent findings [S. Kuwada et al., Science 206, 586–588 (1979); S. Kuwada et al., Neurosci. Abstr. 76 (1979)] in which the interaural phase sensitivity of inferior colliculus neurons has been studied using interaural delays or binaural beats. The cyclic response patterns of some neurons are altered systematically as a function of stimulus off-time and interaural intensity differences. In addition, the response to binaural beat stimuli indicates that some cells are sensitive to the direction of interaural phase change. The model is comprised of two inputs (one from each ear), each with crossed collateral inhibition. The two inputs synapse on a third cell, resulting in four synaptic connections; two inputs are presynaptic inhibitory and two are direct excitatory. At each junction, simple rules about transmitter release, transmitter supply, and changes in membrane potential are made. Manipulation of these parameters allows for simulation of the commonly seen cyclic response pattern as well as some of the unique effects of off-time, interaural intensity differences, and sensitivity to a particular direction of interaural phase change. [Supported by NIH grant NS12732.

Effects of binaural maskers on the subjective laterality of diotic targets

The effect of a simultaneously‐presented dichotic masker on the subjective lateral position of a 500‐Hz diotic target was measured as a function of masker interaural time and intensity differences and target‐to‐masker ratio. The target‐masker complex was presented in temporal alternation with a 500‐Hz binaural pointer tone. Subjects adjusted the interaural intensity difference of this pointer to match the perceived laterality of the partially masked target. It was found that the presence of time‐delayed maskers at intermediate target‐to‐masker ratios tended to attract the perceptual image of the target toward that of the masker, while amplitude‐shifted maskers tended to displace the image of the target away from that of the masker. At target‐to‐masker ratios closer to detection threshold, however, the target tended to be perceived at or near the masker in either case. These dependences are discussed in terms of models of binaural interaction and compared to the ability to discriminate interaural time and intensity differences of the target in the presence of similar maskers. [Supported by NIH.]

Pitch of noise signals: evidence for a "central spectrum".

A review is given of studies dealing with the pitch of signals derived from white Gaussian noise: the pitch of a (monotic) band of noise, the pitch of (dichotic) noise with an interaural phase shift, the monotic and dichotic repetition pitch, and the pitch of two interaurally delayed (uncorrelated) noises. The aim of the paper is to establish the relationship of these pitch phenomena and their connection with the pitch of monaurally presented simple tones (place pitch) and complex tones (periodicity pitch). Therefore, new experiments were performed on the pitch of two noise bands with harmonically related central frequencies, and the pitch of (dichotic) noise with interaural phase shifts at two harmonically related central frequencies, as well as the pitch of band-reject noise. A uniform explanation appears applicable, in correspondence with the theory of the pitch of complex tones, on the basis of the concept of a ’’central spectrum.’’ A simple schematic model of binaural interaction is used to describe the generation of the central spectrum. Pitch data can be accounted for by this model description. The existence region of pitch of dichotic signals is discussed in connection with other binaural functions, like lateralization, binaural beats, and binaural masked level differences.

Effects of interaural frequency differences on the lateralization function.

Lateralization thresholds were determined under conditions of interaural frequency disparity. The lateralization function was independent of the interaural frequency differences employed. Implications for a model of binaural interaction were discussed.