Relations between psychophysical measures of spatial hearing and self-reported spatial-hearing abilities.

OBJECTIVE: The aim of the present study was to investigate how well the virtual psychophysical measures of spatial hearing from the preliminary auditory profile predict self-reported spatial-hearing abilities. DESIGN: Virtual spatial-hearings tests (conducted unaided, via headphones) and a questionnaire were administered in five centres in Germany, the Netherlands, Sweden, and the UK. Correlations and stepwise linear regression models were calculated among a group of hearing-impaired listeners. STUDY SAMPLE: Thirty normal-hearing listeners aged 19-39 years, and 72 hearing-impaired listeners aged 22-91 years with a broad range of hearing losses, including asymmetrical and mixed hearing losses. RESULTS: Several significant correlations (between 0.24 and 0.54) were found between results of virtual psychophysical spatial-hearing tests and self-reported localization abilities. Stepwise linear regression analyses showed that the minimum audible angle (MAA) test was a significant predictor for self-reported localization abilities (5% extra explained variance), and the spatial speech reception threshold (SRT) benefit test for self-reported listening to speech in spatial situations (6% extra explained variance). CONCLUSIONS: The MAA test and spatial SRT benefit test are indicative measures of everyday binaural functioning. The binaural SRT benefit test was not found to predict self-reported spatial-hearing abilities.

A precedence effect in the perception of inter-aural cross correlation.

Does the precedence effect, well known in the field of sound localization or lateralization, also apply to other percepts based on binaural processing? We have compared, with one and the same experimental paradigm, a manifestation of the traditional precedence effect in lateralization with a possible similar effect in the perception of diffuseness or compactness of a sound image. With dichotic headphone stimulation, lateralization was controlled by the inter-aural time delay (IATD), and diffuseness/compactness by the inter-aural cross correlation (IACC). The experimental paradigm rests on the principle of estimating the over-all sensation of a 20-ms noise burst, which was subdivided in two parts, with the relevant dichotic information (IATD or IACC) in the leading part being opposite to that in the trailing part. When each part is 10 ms, it is found that the overall sensation is slightly dominated by the information in the leading part, both for lateralization and for compactness/diffuseness. This dominance of the leading part can be compensated by a certain decrease of its duration and/or amplitude relative to that of the trailing part. It is found that this quantitative measure for the 'strength' of the precedence effect for the present stimulus is essentially the same for IATD and IACC, suggesting that the precedence effect does not apply exclusively to sound localization or lateralization, but to at least one other percept based on binaural processing as well, namely the processing of inter-aural cross correlation.

Stimulus-onset dominance in the perception of binaural information.

With dichotic signals presented by headphone, stimulus-onset dominance (the 'precedence effect') was investigated for various types of binaural-processing-based percepts. The following three dichotic cues were considered: (1) inter-aural time delay (IATD, underlying the lateralization of the sound image), (2) inter-aural level difference (IALD, also underlying lateralization), and (3) inter-aural cross correlation (IACC, underlying the spaciousness of the sound image in terms of broadness/compactness). For all three cases, the degree of stimulus-onset dominance is estimated by one and the same experimental paradigm, which is essentially the same as used by Aoki and Houtgast [Hear. Res. 59, 25-30 (1992)]: When subdividing a brief stimulus in two parts of equal duration, a leading and a trailing part, in which the dichotic cue has opposite values, the over-all sensation is found to be dominated by the cue in the leading part. This dominance can be compensated by shortening the leading part (while keeping total signal duration constant), providing a quantitative measure for the onset dominance. The signals were octave-band filtered noise (center frequencies 500 or 2000 Hz) or 7-kHz low-pass filtered noise, and total signal duration was 5, 10, 20 or 40 ms. The results obtained for the four signal durations have been converted to a weighting function, representing the perceptual weight of the dichotic information as a function of time-after-signal-onset.(ABSTRACT TRUNCATED AT 250 WORDS)

Growth of pulsation threshold of a suppressed tone as a function of its level.

The pulsation threshold (PT) was measured at the frequency of a probe tone in a two-tone stimulus. A suppressor tone was higher in frequency than the probe tone and was fixed in level. As the level of the probe tone was increased, three regions of performance were observed: (1) for probe tone levels below simultaneous masked threshold (SMT), PT was the same as that measured for the suppressor alone, (2) for levels above SMT, PT increased linearly with level, indicating a constant amount of suppression in dB, and (3) for higher levels a recruitment-like phenomenon was observed, in which the PT increased faster than the probe level. The maximum amount of suppression observed was equal to the difference between the PT and SMT for the suppressor alone. One interpretation is that the suppressor reduces excitation on the slopes of its own excitation pattern by the same amount that it reduces the additional excitation from a probe tone. These results are consistent with physiological data, where the amount of suppression is determined by the suppressor and is independent of the level of the probe tone.

The precedence effect for lateralization at low sensation levels.

Using dichotic signals presented by headphone, stimulus onset dominance (the precedence effect) for lateralization at low sensation levels was investigated for five normal hearing subjects. Stimuli were based on 2400-Hz low pass filtered 5-ms noise bursts. We used the paradigm, as described by Aoki and Houtgast (Hear. Res., 59 (1992) 25-30) and Houtgast and Aoki (Hear. Res., 72 (1994) 29-36), in which the stimulus is divided into a leading and a lagging part with opposite lateralization cues (i.e. an interaural time delay of 0.2 ms). The occurrence of onset dominance was investigated by measuring lateral perception of the stimulus, with fixed equal duration of leading and lagging part, while decreasing absolute signal level or adding a filtered white noise with the signal level set at 65 dBA. The dominance of the leading part was quantified by measuring the perceived lateral position of the stimulus as a function of the relative duration of the leading (and thus the lagging) part. This was done at about 45 dB SL without masking noise and also at a signal-to-noise ratio resulting in a sensation level of 10 dB. The occurrence and strength of the precedence effect was found to depend on sensation level, which was decreased either by lowering the signal level or by adding noise. With the present paradigm, besides a decreased lateralization accuracy, a decrease in the precedence effect was found for sensation levels below about 30-40 dB. In daily-life conditions, with a sensation level in noise of typically 10 dB, the onset dominance was still manifest, albeit degraded to some extent.

How we do it: The Dutch functional hearing-screening tests by telephone and internet.

Keypoints * The Dutch National Hearing Test is a reliable and very successful functional hearing-screening test by telephone. An internet version of the National Hearing Test was also implemented. * The National Hearing Test is a fully automatic adaptive speech-in-noise test that uses digit-triplets as speech material. The result of the test is given as 'good,''insufficient,' or 'poor.' * The test by telephone performs better in reaching older subjects, who are more likely to suffer from hearing loss, than the test by internet. * More than 50% of the participants who scored 'insufficient' or 'poor' followed the recommendation to visit a GP, hearing-aid dispenser, ENT specialist or Audiological Center. * The tests contribute to increase the identification and treatment of older hearing-impaired subjects.

Psychophysical measurements relating suppression and combination tones.

Measurement of the combination tone 2f1-f2 with the pulsation-threshold technique yields a significantly lower value than measurement with the cancellation procedure. For three of four subjects, the difference between these two measures equals to amount of suppression produced at 2f1-f2 by the lower primary, f1. It appears that the cancellation procedure overestimates the combination-tone by f1. For a fixed level of f1, the level of the combination tone first increases, then decreases as the level of f2 is raised, for both cancellation and pulsation measures. The level of f2 at which these functions reach a peak is the same level at which f2 begins to suppress the lower primary, f1.

A physical method for measuring speech-transmission quality.

A physical method for measuring the quality of speech-transmission channels has been developed. Essentially, the method represents an extension of the Articulation Index (AI) concept, which was developed mainly to account for distortions in the frequency domain (noise, bandpass-limiting). The underlying concept of the present approach, based on the Modulation Transfer Function (MTF) of a transmission channel, has been adapted to account for nonlinear distortions (peak clipping) as well as for distortions in the time domain (reverberation, echoes, AGC). The resulting index, the Speech-Transmission Index (STI), has been correlated with subjective intelligibility scores obtained on 167 different transmission channels with a wide variety of disturbances. The relative predictive power of the STI, expressed in PB-word score, appeared to be 5%. This accuracy is comparable with results obtained from subjective measurements when about four talkers and four listeners are used. Expressed in terms of signal-to-noise ratio, the accuracy is about 1 dB. Pilot studies have been carried out to evaluate the use of the STI for testing digital-speech transmission channels.

Auditory distance perception in rooms.

The perceived distance of a sound source in a room has been shown to depend on the ratio of the energies of direct and reflected sound. Although this relationship was verified in later studies, the research has never led to a quantitative model. The advent of techniques for the generation of virtual sound sources has made it possible to study distance perception using controlled, deterministic stimuli. Here we present two experiments that make use of such stimuli and we show that a simple model, based on a modified direct-to-reverberant energy ratio, can accurately predict the results and also provide an explanation for the 'auditory horizon' in distance perception. The modification of the ratio consists of the use of an integration time of 6 milliseconds in the calculation of the energy of the direct sound. This time constant seems to be important in spatial hearing-the precedence effect is also based on a similar integration window.

Spectral sharpness and vowel dissimilarity.

The effect of sharpening or smoothing the spectral envelopes of synthetic vowel-like sounds on the dissimilarities perceived among these sounds was investigated by means of triadic comparisons. When a spectral envelope (dB on a log-frequency scale) is considered the sum of a series of sinusoidal spectral modulations (or ripples) of different densities (the ripple spectrum), spectral sharpening or smoothing can be described as an amplification or attenuation of a part of the original ripple spectrum. For a set of nine sounds comprising different degrees of spectral sharpening of a single vowel, the perceived dissimilarities were found to be dominated by a specific part of the ripple spectrum, i.e., by spectral modulations with a density of about 2 ripples/oct. The possible role of lateral suppression in relation to this dominant region is discussed. For a set of 18 sounds comprising six vowels, each in three different versions (sharpened, normal, or smoothed), the dissimilarities were found to be determined mainly by the global shape of the spectral envelopes, i.e., by spectral modulations up to about 1.5-2 ripples/oct. Details of the spectral envelope (including the region of 2 ripples/oct where lateral suppression is effective) appear to be of minor influence on vowel dissimilarities.

Measuring the threshold for speech reception by adaptive variation of the signal bandwidth. II. Hearing-impaired listeners.

In a previous study [Noordhoek et al., J. Acoust. Soc. Am. 105, 2895-2902 (1999)], an adaptive test was developed to determine the speech-reception bandwidth threshold (SRBT), i.e., the width of a speech band around 1 kHz required for a 50% intelligibility score. In this test, the band-filtered speech is presented in complementary bandstop-filtered noise. In the present study, the performance of 34 hearing-impaired listeners was measured on this SRBT test and on more common SRT (speech-reception threshold) tests, namely the SRT in quiet, the standard SRT in noise (standard speech spectrum), and the spectrally adapted SRT in noise (fitted to the individual's dynamic range). The aim was to investigate to what extent the performance on these tests could be explained simply from audibility, as estimated with the SII (speech intelligibility index) model, or require the assumption of suprathreshold deficits. For most listeners, an elevated SRT in quiet or an elevated standard SRT in noise could be explained on the basis of audibility. For the spectrally adapted SRT in noise, and especially for the SRBT, the data of most listeners could not be explained from audibility, suggesting that the effects of suprathreshold deficits may be present. Possibly, such a deficit is an increased downward spread of masking.

Measuring the threshold for speech reception by adaptive variation of the signal bandwidth. I. Normal-hearing listeners.

An adaptive test has been developed to determine the minimum bandwidth of speech that a listener needs to reach 50% intelligibility. Measuring this speech-reception bandwidth threshold (SRBT), in addition to the more common speech-reception threshold (SRT) in noise, may be useful in investigating the factors underlying impaired suprathreshold speech perception. Speech was bandpass filtered (center frequency: 1 kHz) and complementary bandstop filtered noise was added. To obtain reference values, the SRBT was measured in 12 normal-hearing listeners at four sound-pressure levels, in combination with three overall spectral tilts. Plotting SRBT as a function of sound-pressure level resulted in U-shaped curves. The most narrow SRBT (1.4 octave) was obtained at an A-weighted sound-pressure level of 55 dB. The required bandwidth increases with increasing level, probably due to upward spread of masking. At a lower level (40 dBA) listeners also need a broader band, because parts of the speech signal will be below threshold. The SII (Speech Intelligibility Index) model reasonably predicts the data, although it seems to underestimate upward spread of masking.

Compression and expansion of the temporal envelope: evaluation of speech intelligibility and sound quality.

Sensorineural hearing loss is accompanied by loudness recruitment, a steeper-than-normal rise of perceived loudness with presentation level. To compensate for this abnormality, amplitude compression is often applied (e.g., in a hearing aid). Alternatively, since speech intelligibility has been modeled as the perception of fast energy fluctuations, enlarging these (by means of expansion) may improve speech intelligibility. Still, even if these signal-processing techniques prove useful in terms of speech intelligibility, practical application might be hindered by unacceptably low sound quality. Therefore, both speech intelligibility and sound quality were evaluated for syllabic compression and expansion of the temporal envelope. Speech intelligibility was evaluated with an adaptive procedure, based on short everyday sentences either in noise or with a competing speaker. Sound quality was measured by means of a rating-scale procedure, for both speech and music. In a systematic setup, both the ratio of compression or expansion and the number of independent processing bands were varied. Individual hearing thresholds were compensated for by a listener-specific filter and amplification. Both listeners with normal hearing and listeners with sensorineural hearing impairment participated as paid volunteers. The results show that, on average, both compression and expansion fail to show better speech intelligibility or sound quality than linear amplification.

Intensity discrimination of Gaussian-windowed tones: indications for the shape of the auditory frequency-time window.

The just-noticeable difference in intensity jnd(I) was measured for 1-kHz tones with a Gaussian-shaped envelope as a function of their spectro-temporal shape. The stimuli, with constant energy and a constant product of bandwidth and duration, ranged from a long-duration narrow-band "tone" to a short-duration broadband "click." The jnd(I) was measured in three normal-hearing listeners at sensation levels of 0, 10, 20, and 30 dB in 35 dB(A) SPL pink noise. At intermediate sensation levels, jnd(I) depends on the spectro-temporal shape: at the extreme shapes (tones and clicks), intensity discrimination performance is best, whereas at intermediate shapes the jnd(I) is larger. Similar results are observed at a higher overall sound level, and at a higher carrier frequency. The maximum jnd(I) is observed for stimuli with an effective bandwidth of about 1/3 octave and an effective duration of 4 ms at 1 kHz (1 ms at 4 kHz). A generalized multiple-window model is proposed that assumes that the spectro-temporal domain is partitioned into "internal" auditory frequency-time windows. The model predicts that intensity discrimination thresholds depend upon the number of windows excited by a signal: jnd(I) is largest for stimuli covering one window.

Preference judgments of artificial processed and hearing-aid transduced speech.

In order to assess the relative importance of various signal processing algorithms and distortions on hearing-aid preference, male and female speech was manipulated in a number of ways and subsequently presented to normal-hearing and hearing-impaired subjects (the latter having a mild sensorineural high-frequency hearing loss). Signal manipulations were artificial (e.g., filtering, compression, peak clipping, or adding noise) or were actual dummy-head recordings of five different hearing aids. Listeners judged the sounds in a pairwise-comparison format. Their task was to indicate the "hearing aid" they would prefer assuming they had to wear it all day. The data were analyzed with multidimensional scaling techniques; Principal Components Analysis revealed that the first two dimensions on which preference judgments were based, can be interpreted as (1) intelligibility or clarity, and (2) distinction between signal distortion and added background distortion. Furthermore, the results showed that normal-hearing subjects generally preferred the original signal, whereas hearing-impaired subjects were inclined to choose the signals with a high-frequency emphasis. Severe band-pass filtering or low-frequency emphasis were disliked, as was to be expected. Surprisingly, however, a soft background noise (S/N ratio of 25 dB) was often among the least preferred of all signals. The differences in preference between the five hearing aids were small, but consistent. For hearing-impaired subjects, hearing-aid ordering could be accounted for by the amount of low-frequency cutoff; for normal-hearing subjects both high- and low-frequency cutoff played a role. Results of a retest experiment with normal-hearing subjects, about one year later, showed that subjects' criteria remain remarkably stable.

Peaks in the frequency response of hearing aids: evaluation of the effects on speech intelligibility and sound quality.

In a series of experiments, we introduced peaks of 10, 20, and 30 dB, in various combinations, onto a smooth reference frequency response. For each of the conditions, we evaluated speech intelligibility in noise, using a test as developed by Plomp and Mimpen (1979), and sound quality (for both speech and music), using a rating-scale procedure. We performed the experiments with 26 listeners with sensorineurally impaired hearing and 10 listeners with normal hearing. Signal processing was accomplished digitally; for each listener, the stimuli were filtered and subsequently amplified so that the average speech spectrum was well above the threshold of hearing at all frequencies. The results show that, as a result of the introduction of peaks onto the frequency response, speech intelligibility is affected more for the listeners with impaired hearing than for those with normal hearing. Sound-quality judgments tend to be less different between the listener groups. Conditions with 30-dB peaks especially show serious effects on both speech intelligibility and sound quality.

Effect of temporal modulation reduction on spectral contrasts in speech.

In this paper the effect of temporal modulation reduction on spectral contrasts is investigated. First, a spectral modulation transfer function (SMTF) is presented as a method to measure the transfer of spectral ripples (sinusoidal periods/oct) in the short-time spectral envelope by comparing the spectral modulation depth of original and processed speech fragments. Measuring the SMTF for speech subjected to uniform reduction of the temporal modulation depth (i.e., modulation-frequency-independent reduction) in 24 1/4-oct bands showed an almost equal uniform reduction of the spectral modulations. Furthermore, the SMTF was used to measure the reduction of spectral contrasts associated with low-pass and high-pass temporal-envelope filtering [Drullman et al., J. Acoust. Soc. Am.95, 1053-1064 and 2670-2680 (1994a, b)]. For a perceptual evaluation, sentences were processed to reduce spectral contrasts and the speech-reception threshold (SRT) in noise was measured with ten normal-hearing subjects. Comparison of the results with those obtained previously after temporal-envelope filtering revealed that the SRT-effect of temporal high-pass filtering can be completely accounted for by the associated reduction of spectral contrasts. However, this relationship cannot be demonstrated conclusively in the case of temporal low-pass filtering.

Spectro-temporal integration in signal detection.

This paper is concerned with aspects of temporal integration and across-frequency integration in signal detection. Previous experiments on the detection of brief broadband signals (clicks) in continuous broadband noise revealed efficient spectral integration. The extent to which this effect is restricted to a critical time window was investigated by manipulating the temporal relations among the signal components in different frequency regions. In a typical experiment, the signal consists of nine brief Gaussian-shaped tone pulses, equally distributed at 1/3-oct intervals, each with a spectral width of about 1/3 oct, and each equally detectable in white noise. In the synchronized condition (i.e. coinciding peaks of the nine Gaussian envelopes), the detection threshold is reached when the levels of the nine individual tone pulses are about 8 dB below their individual threshold levels (efficient spectral integration). When the signal is progressively desynchronized (i.e. noncoinciding peaks of the Gaussian envelopes), detection threshold is found to increase. This suggests that efficient spectral integration in signal detection is confined to a narrow time window, with a typical value of 30 ms. Similar experiments were performed with respect to the efficiency of temporal integration. For constant-duration signals (100 ms), the detection threshold is found to increase when progressively widening signal bandwidth. The data indicate that the efficient temporal integration in signal detection is confined to a narrow frequency window, which, not surprisingly, corresponds to the critical bandwidth.

Efficient across-frequency integration in short-signal detection.

A series of experiments was performed on the influence of bandwidth on the masked threshold of brief deterministic signals in continuous broadband noise. The signal bandwidth is quantified by the number (n) of constituent 1/3-oct bands. For n increasing from 1 to typically 9, the masked-threshold level in the constituent 1/3-oct bands is found to decrease by 8 log(n). This integration rule is obtained when each of the 1/3-oct bands covered by the signal equally contributes to detection, i.e., that, for each of these 1/3-oct bands, the difference between signal level and the individual masked-threshold level is the same. It was found that this integration rule also applies to noncontiguous signal spectra and that it remains intact over a broad range of masker levels. Commonly, the masked threshold of compound signals (for instance, n frequency components with a spacing of typically 1/3 oct), relative to the masked threshold of single-component signals, has been described by a 5 log(n) integration rule. However, this rule was obtained for signal durations of typically 100 ms or more. For the present brief signals (typically 10 ms or less), the across-frequency integration is found to be more effective.

Effects of degradation of intensity, time, or frequency content on speech intelligibility for normal-hearing and hearing-impaired listeners.

Many hearing-impaired listeners suffer from distorted auditory processing capabilities. This study examines which aspects of auditory coding (i.e., intensity, time, or frequency) are distorted and how this affects speech perception. The distortion-sensitivity model is used: The effect of distorted auditory coding of a speech signal is simulated by an artificial distortion, and the sensitivity of speech intelligibility to this artificial distortion is compared for normal-hearing and hearing-impaired listeners. Stimuli (speech plus noise) are wavelet coded using a complex sinusoidal carrier with a Gaussian envelope (1/4 octave bandwidth). Intensity information is distorted by multiplying the modulus of each wavelet coefficient by a random factor. Temporal and spectral information are distorted by randomly shifting the wavelet positions along the temporal or spectral axis, respectively. Measured were (1) detection thresholds for each type of distortion, and (2) speech-reception thresholds for various degrees of distortion. For spectral distortion, hearing-impaired listeners showed increased detection thresholds and were also less sensitive to the distortion with respect to speech perception. For intensity and temporal distortion, this was not observed. Results indicate that a distorted coding of spectral information may be an important factor underlying reduced speech intelligibility for the hearing impaired.