Brian R. Glasberg

Derivation of auditory filter shapes from notched-noise data

A well established method for estimating the shape of the auditory filter is based on the measurement of the threshold of a sinusoidal signal in a notched-noise masker, as a function of notch width. To measure the asymmetry of the filter, the notch has to be placed both symmetrically and asymmetrically about the signal frequency. In previous work several simplifying assumptions and approximations were made in deriving auditory filter shapes from the data. In this paper we describe modifications to the fitting procedure which allow more accurate derivations. These include: 1) taking into account changes in filter bandwidth with centre frequency when allowing for the effects of off-frequency listening; 2) correcting for the non-flat frequency response of the earphone; 3) correcting for the transmission characteristics of the outer and middle ear; 4) limiting the amount by which the centre frequency of the filter can shift in order to maximise the signal-to-masker ratio. In many cases, these modifications result in only small changes to the derived filter shape. However, at very high and very low centre frequencies and for hearing-impaired subjects the differences can be substantial. It is also shown that filter shapes derived from data where the notch is always placed symmetrically about the signal frequency can be seriously in error when the underlying filter is markedly asymmetric. New formulae are suggested describing the variation of the auditory filter with frequency and level. The implication of the results for the calculation of excitation patterns are discussed and a modified procedure is proposed. The appendix list FORTRAN computer programs for deriving auditory filter shapes from notched-noise data and for calculating excitation patterns. The first program can readily be modified so as to derive auditory filter shapes from data obtained with other types of maskers, such as rippled noise.

Suggested formulae for calculating auditory-filter bandwidths and excitation patterns

Recent estimates of auditory-filter shape are used to derive a simple formula relating the equivalent rectangular bandwidth (ERB) of the auditory filter to center frequency. The value of the auditory-filter bandwidth continues to decrease as center frequency decreases below 500 Hz. A formula is also given relating ERB-rate to frequency. Finally, a method is described for calculating excitation patterns from filter shapes.

A test for the diagnosis of dead regions in the cochlea.

Abstract Hearing impairment may sometimes be associated with complete loss of inner hair cells (IHCs) over a certain region of the basilar membrane. We call this a ‘dead region’. Amplification (using a hearing aid) over a frequency range corresponding to a dead region may not be beneficial and may even impair speech intelligibility. However, diagnosis of dead regions is not easily done from the audiogram. This paper reports the design and evaluation of a method for detecting and delimiting dead regions. A noise, called ‘threshold equalizing noise’ (TEN), was spectrally shaped so that, for normally hearing subjects, it would give equal masked thresholds for pure tone signals at all frequencies within the range 250–10 000 Hz. Its level is specified as the level in a one-ERB (132 Hz) wide band centred at 1000 Hz. Measurements obtained from 22 normal-hearing subjects and TEN levels of 30, 50 and 70 dB/ERB confirmed that the signal level at masked threshold was approximately equal to the noise level/ERB and was almost independent of signal frequency. Masked thresholds were measured for 20 ears of 14 subjects with sensorineural hearing loss, using TEN levels of 30, 50 and 70 dB/ERB. Psychophysical tuning curves (PTCs) were measured for the same subjects. When there are surviving IHCs corresponding to a frequency region with elevated absolute thresholds, a signal in that frequency region is detected via IHCs with characteristic frequencies (CFs) close to that region. In such a case, threshold in the TEN is close to that for normal-hearing listeners, provided that the noise intensity is sufficient to produce significant masking. Also, the tip of the PTC lies close to the signal frequency. When a dead region is present, the signal is detected via IHCs with CFs different from that of the signal frequency. In such a case, threshold in the TEN is markedly higher than normal, and the tip of the PTC is shifted away from the signal frequency. Generally, there was a very good correspondence between the results obtained using the TEN and the PTCs. We conclude that the measurement of masked thresholds in TEN provides a quick and simple method for the diagnosis of dead regions.

Auditory filter shapes in subjects with unilateral and bilateral cochlear impairments

The shape of the auditory filter was estimated at three center frequencies, 0.5, 1.0, and 2.0 kHz, for five subjects with unilateral cochlear impairments. Additional measurements were made at 1.0 kHz using one subject with a unilateral impairment and six subjects with bilateral impairments. Subjects were chosen who had thresholds in the impaired ears which were relatively flat as a function of frequency and ranged from 15 to 70 dB HL. The filter shapes were estimated by measuring thresholds for sinusoidal signals (frequency f) in the presence of two bands of noise, 0.4 f wide, one above and one below f. The spectrum level of the noise was 50 dB (re: 20 mu Pa) and the noise bands were placed both symmetrically and asymmetrically about the signal frequency. The deviation of the nearer edge of each noise band from f varied from 0.0 to 0.8 f. For the normal ears, the filters were markedly asymmetric for center frequencies of 1.0 and 2.0 kHz, the high-frequency branch being steeper. At 0.5 kHz, the filters were more symmetric. For the impaired ears, the filter shapes varied considerably from one subject to another. For most subjects, the lower branch of the filter was much less steep than normal. The upper branch was often less steep than normal, but a few subjects showed a near normal upper branch. For the subjects with unilateral impairments, the equivalent rectangular bandwidth of the filter was always greater for the impaired ear than for the normal ear at each center frequency. For three subjects at 0.5 kHz and one subject at 1.0 kHz, the filter had too little selectivity for its shape to be determined.

Formulae describing frequency selectivity as a function of frequency and level, and their use in calculating excitation patterns

The auditory filter may be considered as a weighting function representing frequency selectivity at a particular centre frequency. Its shape can be derived using the power-spectrum model of masking which assumes: (1) in detecting a signal in a masker the observer uses the single auditory filter giving the highest signal-to-masker ratio; (2) threshold corresponds to a fixed signal-to-masker ratio at the output of that filter. Factors influencing the choice of a masker to measure the auditory filter shape are discussed. Narrow-band maskers are unsuitable for this purpose, since they violate the assumptions of the power-spectrum model. A method using a notched-noise masker is recommended, and typical results using that method are presented. The variation of the auditory filter shape with centre frequency and with level, and the relationship of the auditory filter shape and the excitation pattern are described. A method of calculating the excitation pattern of any sound as a function of level is presented, and examples and applications are given. The appendix gives a Fortran program for calculating excitation patterns.

The shape of the ear’s temporal window

This article examines the idea that the temporal resolution of the auditory system can be modeled using a temporal window (an intensity weighting function) analogous to the auditory filter measured in the frequency domain. To estimate the shape of the hypothetical temporal window, threshold was measured for a brief sinusoidal signal presented in a temporal gap between two bursts of noise. The duration of the gap was systematically varied and the signal was placed both symmetrically and asymmetrically within the gap. The data were analyzed by assuming that the temporal window had the form of a simple mathematical expression with a small number of free parameters. The values of the parameters were adjusted to give the best fit to the data. The analysis assumed that, for each condition, the temporal window was centered at the time giving the highest signal-to-masker ratio, and that threshold corresponded to a fixed ratio of signal energy to masker energy at the output of the window. The data were fitted well by modeling each side of the window as the sum of two rounded-exponential functions. The window was highly asymmetric, having a shallower slope for times before the center than for times after. The equivalent rectangular duration (ERD) of the window was typically about 8 ms. The ERD increased slightly when the masker level was decreased, but did not differ significantly for signal frequencies of 500 and 2000 Hz. The temporal-window model successfully accounts for the data from a variety of experiments measuring temporal resolution. However, it fails to predict certain aspects of forward masking and of the detection of amplitude modulation at high rates.

Relative dominance of individual partials in determining the pitch of complex tones

These experiments were conducted to determine the dominance of each partial in determining the residue pitch of a complex tone. Subjects were required to make pitch matches to a complex tone which had one partial slightly mistuned from its ‘‘correct’’ harmonic value. The shift in residue pitch was measured as a function of the frequency shift of the harmonic, for each harmonic in turn. For mistunings up to ±2%–3% the shift in residue pitch was approximately a linear function of the shift in the harmonic, but for greater mistunings the shift in residue pitch was reduced. The degree to which a given harmonic can influence residue pitch gives a measure of the dominance of that harmonic. The dominant harmonics were always contained within the lowest six harmonics (for fundamental frequencies of 100, 200, and 400 Hz), but there were marked individual differences in the exact distribution of dominance across harmonics. The level of a harmonic relative to adjacent harmonics can have a significant effect on its d...

A revised model of loudness perception applied to cochlear hearing loss

We previously described a model for loudness perception for people with cochlear hearing loss. However, that model is incompatible with our most recent and most satisfactory model of loudness for normal hearing. Here, we describe a loudness model that is applicable to both normal and impaired hearing. In contrast to our earlier model for impaired hearing, the new model correctly predicts: (1) that a sound at absolute threshold has a small but finite loudness; (2) that, for levels very close to the absolute threshold, the rate of growth of loudness is similar for normal ears and ears with cochlear hearing loss; (3) the relation between monaural and binaural threshold and loudness; (4) recent measures of equal-loudness contours. Like the earlier model, the new model can account for the loudness recruitment and reduced loudness summation that are typically associated with cochlear hearing loss.

Gap detection and masking in hearing‐impaired and normal‐hearing subjects

Subjects with cochlear impairments often show reduced temporal resolution as measured in gap-detection tasks. The primary goals of these experiments were: to assess the extent to which the enlarged gap thresholds can be explained by elevations in absolute threshold; and to determine whether the large gap thresholds can be explained by the same processes that lead to a slower-than-normal recovery from forward masking. In experiment I gap thresholds were measured for nine unilaterally and eight bilaterally impaired subjects, using bandlimited noise stimuli centered at 0.5, 1.0, and 2.0 kHz. Gap thresholds were usually larger for the impaired ears, even when the comparisons were made at equal sensation levels (SLs). Gap thresholds tended to increase with increasing absolute threshold, but the scatter of gap thresholds was large for a given degree of hearing loss. In experiment II threshold was measured as a function of the delay between the onset of a 210-ms masker and the onset of a 10-ms signal in both simultaneous- and forward-masking conditions. The signal frequency was equal to the center frequency of the bandlimited noise masker, which was 0.5, 1.0, or 2.0 kHz. Five subjects with unilateral cochlear impairments, two subjects with bilateral impairments, and two normal subjects were tested. The rate of recovery from forward masking, particularly the initial rate, was usually slower for the impaired ears, even when the maskers were presented at equal SLs. Large gap thresholds tended to be associated with slow rates of recovery from forward masking.

Frequency selectivity as a function of level and frequency measured with uniformly exciting notched noise.

Thresholds for detecting sinusoidal signals were measured as a function of the spectral width of a notch in a noise masker. The notch was positioned both symmetrically and asymmetrically around the signal frequency. The noise was designed to create equal excitation per ERB within its passbands (uniformly exciting noise), after allowing for the transfer function of the headphone and the middle ear. For a signal frequency of 250 Hz, the level per ERB ranged from 35 to 80 dB in 15-dB steps. For signal frequencies of 500, 1,000, 2,000, and 4,000 Hz, the level per ERB ranged from 40 to 70 dB per ERB in 15-dB steps. Auditory filter shapes were derived from the data by modeling the auditory filter as the sum of a sharply tuned tip filter and a broader tail filter. The gain of the tip filter was assumed to be a function of level. The shape of the tip filter and the gain and shape of the tail filter were assumed to be level independent. The data for all levels were fitted simultaneously. The data were fitted best when the gain of the tip filter was assumed to be a function of the signal level (as opposed to the masker level per ERB). The filter shapes showed a level dependence that qualitatively resembled the level dependence of filtering on the basilar membrane. The maximum gain of the tip filter tended to increase with increasing center frequency up to 1 kHz, but to remain roughly constant for higher frequencies.