MORE RELATED ABSTRACTS:

Speech Segregation Based on Sound Localization - Roman, Wang  

....plausible network [13] This mechanism accounts for the lateral displacement of the auditory event from the median plane when an ITD is present. Mechanisms additional to the cross correlation model have been proposed to simulate auditory event localization based on both on ITD and IID [4] 5] [8]. Models of binaural hearing have already been used for sound source separation [9] 6] The main underlying observation in most of the existing models is that the auditory event corresponding to a desired sound source undergoes systematic changes due to the interfering noise. Our model attempts ....

....rectified in order to simulate the firing probabilities of nerve fibers. Saturation effects are modeled by taking the square root of the signal. 2. 3 Azimuth Locator Current models of azimuth localization use, as a starting point, Jeffress s cross correlation mechanism [13] 4] 5] 6] [8], 9] Cross correlation provides excellent time delay estimation for broadband signals and narrowband stimuli in the low frequency range. However, for periodic waveforms it can present ambiguous peaks at intervals of the fundamental frequency. In our model the crosscorrelation is implemented by ....

Multi-Microphone Correlation-Based Processing for Robust.. - Sullivan (1996)   (10 citations)  (Correct)

....with respect to the other. He also hypothesized that the same mechanism could be used for localizing frequencies above 1500 Hz which are detected by level differences between the two ears (interaural level differences, ILDs) by looking at the level differences at the outputs of the smaller fibers. Stern and Colburn [1978] elaborated on the Jeffress model. Their model describes a mechanism for generating an estimate of the internal cross correlation function which the nerve fibers calculate. The model describes the nerve firing rates by statistically independent Poisson processes. They also extended their model by ....

A simulation of vowel segregation based on across-channel.. - Ellis (1993)   (Correct)

....estimate of equation (10) with the advantage that it permits the use of small integers for each element in the array. We note in passing the similarity between this matrix detecting co occurrences of firings in different frequency channels to the binaural co incidence detector of Colburn (1977), which detects consistent timing skew between channels of the same center frequency, but from opposing ears, as a factor contributing to perceived spatial origin. 0 relative timing firing channel other channels timeskewed event 3Dhist 1993oct05 figure 6: The three dimensional histogram for ....

Estimating Azimuth And Elevation From Interaural Differences - Keith Martin Perceptual (1995)   (Correct)

....provide a small time window with a locally high signal to noise ratio for evaluating interaural differences. Psychologically, there is evidence that interaural differences are strongly weighted by human listeners at onsets. This has been demonstrated in experiments related to the precedence effect [7]. At present, an extremely simple method of onset detection has been implemented. In each frequency band, the two envelope signals are summed, and local maxima are identified. A simple suppression mechanism has been used to model the precedence effect, whereby selecting an onset causes envelope ....

....Differences. In the current model, the IID is simply formed by the weighted log ratio of the envelope signals. The weighting is accomplished by a window function, which takes the form w(t) At exp( Gammat= The effective width of the window used is 2 3 ms, as suggestedby the results in [7], and the peak of the window function is centered at each detected onset. This weighting function is an important part of the precedenceeffect model, allowing suppressed onsets to have some small influence on the IID estimated by the model. The IPD is estimated by a running cross correlator ....

Echo Suppression In A Computational Model Of The Precedence.. - Keith Martin   (Correct)

.... 1 ms IBD Dt 2 Dt 2 Dt 2 (a) b) c) LEFT RIGHT Segment Delay Dt 2 50 TIME TIME (ms) first second third Figure 1: Three different stimulus paradigms for psychophysical investigation of the precedence effect: a) binaural click pair (after [1] b) Zurek s 3AFC binaural noise burst pairs (after [4]) and (c) Zurek s sub burst paradigm (after [4] See text for details. 4] This suppression been shown to operate independently in different frequency bands [5] and it seems to be one of the primary components of the precedence effect, though it does not account for the active processes ....

.... Segment Delay Dt 2 50 TIME TIME (ms) first second third Figure 1: Three different stimulus paradigms for psychophysical investigation of the precedence effect: a) binaural click pair (after [1] b) Zurek s 3AFC binaural noise burst pairs (after [4] and (c) Zurek s sub burst paradigm (after [4]) See text for details. 4] This suppression been shown to operate independently in different frequency bands [5] and it seems to be one of the primary components of the precedence effect, though it does not account for the active processes found in more recent psychophysical data (c.f. 5, ....

A Computational Model of Spatial Hearing - Martin (1995)   (4 citations)  (Correct)

....shown that transients (i.e. sharp changes in energy level) must be present in the sound signal for the precedence effect to occur. It is not clear how the precedence effect operates across different frequency bands [40] Quantitative measures of the precedence effect are presented by Zurek in [39], and a model has been proposed which might explain the experimental evidence [40] The basic form of the model is shown in Figure 2 2. Inhibition Perception Time Average Onset Detection Instantaneous Localization Lateralization L R Figure 2 2: A possible model for the precedence effect, as ....

....in free field. We also present a brief analysis of the localization blur exhibited by the model, based on the broad band localization experiment and on perceptual distance maps. Finally, we report the results of an experiment in which stimuli used by Zurek to quantify the precedence effect ([39]) were presented to the model. All of the results in this chapter were calculated with the same variance parameters in the position estimator. We assume that the standard deviation of the IID noise is 0:65 dB and that the standard deviations of the IPD and IED noises are 0:05 ms and 0:1 ms ....