Search

The objective of this paper was to elucidate the relation between psychophysical lateralization and the neural generators of the corresponding auditory evoked potentials. Auditory brain stem responses to binaural click stimuli with different interaural time- and level differences were obtained in 12 subjects by means of multi-channel EEG recording. Data were modeled by equivalent current dipoles representing the generating sources in the brain. A generalized maximum-likelihood method was used to solve the inverse problem, taking into account the noise covariance matrix of the data. The quality of the fit was assessed by computing the goodness-of-fit as the outcome of a chi 2-test. This measure was advantageous compared to the conventionally employed residual variance. At the latency of Jewett wave V, there was a systematic variation of the moment of a rotating dipole with the lateralization of the stimulus. Dipole moment trajectories of stimuli with similar lateralization were similar. A sign reversal of the interaural differences resulted in a mirrored trajectory. Centrally-perceived stimuli corresponded to dipoles with the largest vertical components. With increasing lateralization, the vertical component of the moment decreased, while the horizontal components increased. The similarity of trajectories inducted by the same lateralization show that interaural time- and level differences are not processed independently. The present data support the notion that directional information is already extracted and represented at the level of the brain stem.

Rising chirps that compensate for the dispersion of the travelling wave on the basilar membrane evoke larger monaural brainstem responses than clicks. In order to test if a similar effect applies for the early processing stages of binaural information, monaurally and binaurally evoked auditory brainstem responses were recorded for clicks and chirps for levels from 10 to 60 dB nHL in steps of 10 dB. Ten thousand sweeps were collected for every stimulus condition from 10 normal hearing subjects. Wave V amplitudes are significantly larger for chirps than for clicks for all conditions. The amplitude of the binaural difference potential, DP1-DN1, is significantly larger for chirps at the levels 30 and 40 dB nHL. Both the binaurally evoked potential and the binaural difference potential exhibit steeper growth functions for chirps than for clicks for levels up to 40 dB nHL. For higher stimulation levels the chirp responses saturate approaching the click evoked amplitude. For both stimuli the latency of DP1 is shorter than the latency of the binaural wave V, which in turn is shorter than the latency of DN1. The amplitude ratio of the binaural difference potential to the binaural response is independent of stimulus level for clicks and chirps. A possible interpretation is that with click stimulation predominantly binaural interaction from high frequency regions is seen which is compatible with a processing by contralateral inhibitory and ipsilateral excitatory (IE) cells. Contributions from low frequencies are negligible since the responses from low frequencies are not synchronized for clicks. The improved synchronization at lower frequencies using chirp stimuli yields contributions from both low and high frequency neurons enlarging the amplitudes of the binaural responses as well as the binaural difference potential. Since the constant amplitude ratio of the binaural difference potential to the binaural response makes contralateral and ipsilateral excitatory interaction improbable, binaural interaction at low frequencies is presumably also of the IE type. Another conclusion of this study is that the chirp stimuli employed here are better suited for auditory brainstem responses and binaural difference potentials than click stimuli since they exhibit higher amplitudes and a better signal-to-noise ratio.

The minimum standard deviations achievable for concurrent estimates of thresholds and psychometric function slopes as well as the optimal target values for adaptive procedures are calculated as functions of stimulus level and track length on the basis of the binomial theory. The optimum pair of targets for a concurrent estimate is found at the correct response probabilities p1 = 0.19 and p2 = 0.81 for the logistic psychometric function. An adaptive procedure that converges at these optimal targets is introduced and tested with Monte Carlo simulations. The efficiency increases rapidly when each subject's response consists of more than one statistically independent Bernoulli trial. Sentence intelligibility tests provide more than one Bernoulli trial per sentence when each word is scored separately. The number of within-sentence trials can be quantified by the j factor [Boothroyd and Nittrouer, J. Acoust. Soc. Am. 84, 101-114 (1988)]. The adaptive procedure was evaluated with 10 normal-hearing and 11 hearing-impaired listeners using two German sentence tests that differ in j factors. The expected advantage of the sentence test with the higher j factor was not observed, possibly due to training effects. Hence, the number of sentences required for a reliable speech reception threshold (approximately 1 dB standard deviation) concurrently with a slope estimate (approximately 20%-30% relative standard deviation) is at least N = 30 if word scoring for short, meaningful sentences (j approximately 2) is performed.

Loudness was measured as a function of signal bandwidth for 10-, 100-, and 1000-ms-long signals. The test and reference signals were bandpass-filtered noise spectrally centered at 2 kHz. The bandwidth of the test signal was varied from 200 to 6400 Hz. The reference signal had a bandwidth of 3200 Hz. The reference levels were 45, 55, and 65 dB SPL. The level to produce equal loudness was measured with an adaptive, two-interval, two-alternative forced-choice procedure. A loudness matching procedure was used, where the tracks for all signal pairs to be compared were interleaved. Mean results for nine normal-hearing subjects showed that the magnitude of spectral loudness summation depends on signal duration. For all reference levels, a 6- to 8-dB larger level difference between equally loud signals with the smallest (delta f = 200 Hz) and largest (delta f = 6400 Hz) bandwidth is found for 10-ms-long signals than for the 1000-ms-long signals. The duration effect slightly decreases with increasing reference loudness. As a consequence, loudness models should include a duration-dependent compression stage. Alternatively, if a fixed loudness ratio between signals of different duration is assumed, this loudness ratio should depend on the signal spectrum.

The dependence of binaurally evoked auditory brain stem responses and the binaural difference potential on simultaneously presented interaural time and level differences is investigated in order to assess the representation of stimulus lateralization in the brain stem. Auditory brain stem responses to binaural click stimuli with all combinations of three interaural time and three interaural level differences were recorded from 12 subjects and 4 channels. The latency of Jewett wave V is shortest for zero interaural time difference and longest for the trading stimuli. The amplitude of wave V is largest for centrally perceived stimuli, i.e., the diotic and trading stimuli, and smallest for the most laterally perceived stimuli. The latency of the most prominent peak of the binaural difference potential DN1 mainly depends on the interaural time difference. The amplitude of the components of the binaural difference potential, DP1-DN1, depends similarly on stimulus conditions as wave V amplitude in the case of the binaural stimuli: smallest amplitudes are found for the most lateral stimuli and largest amplitudes for central stimuli. The results demonstrate that interaural level and time differences are not processed independently. This supports the hypothesis that directional information in humans is already extracted and represented at the level of the brain stem.

This study examines auditory brainstem responses (ABR) elicited by rising frequency chirps. The time course of frequency change for the chirp theoretically produces simultaneous displacement maxima by compensating for travel-time differences along the cochlear partition. This broadband chirp was derived on the basis of a linear cochlea model [de Boer, "Auditory physics. Physical principles in hearing theory I," Phys. Rep. 62, 87-174 (1980)]. Responses elicited by the broadband chirp show a larger wave-V amplitude than do click-evoked responses for most stimulation levels tested. This result is in contrast to the general hypothesis that the ABR is an electrophysiological event most effectively evoked by the onset or offset of an acoustic stimulus, and unaffected by further stimulation. The use of this rising frequency chirp enables the inclusion of activity from lower frequency regions, whereas with a click, synchrony is decreased in accordance with decreasing traveling velocity in the apical region. The use of a temporally reversed (falling) chirp leads to a further decrease in synchrony as reflected in ABR responses that are smaller than those from a click. These results are compatible with earlier experimental results from recordings of compound action potentials (CAP) [Shore and Nuttall, "High synchrony compound action potentials evoked by rising frequency-swept tonebursts," J. Acoust. Soc. Am. 78, 1286-1295 (1985)] reflecting activity at the level of the auditory nerve. Since the ABR components considered here presumably reflect neural response from the brainstem, the effect of an optimized synchronization at the peripheral level can also be observed at the brainstem level. The rising chirp may therefore be of clinical use in assessing the integrity of the entire peripheral organ and not just its basal end.

Critical experiments were performed in order to validate the two-source hypothesis of distortion product otoacoustic emissions (DPOAE) generation. Measurements of the spectral fine structure of DPOAE in response to stimulation with two sinusoids have been performed with normal-hearing subjects. The dependence of fine-structure patterns on the frequency ratio f2/f1 was investigated by changing f1 or f2 only (fixed f2 or fixed f1 paradigm, respectively), and by changing both primaries at a fixed ratio and looking at different order DPOAE. When f2/f1 is varied in the fixed ratio paradigm, the patterns of 2 f1-f2 fine structure vary considerably more if plotted as a function of f2 than as a function of fDP. Different order distortion products located at the same characteristic place on the basilar membrane (BM) show similar patterns for both, the fixed-f2 and fDP paradigms. Fluctuations in DPOAE level up to 20 dB can be observed. In contrast, the results from a fixed-fDP paradigm do not show any fine structure but only an overall dependence of DP level on the frequency ratio, with a maximum for 2f1-f2 at f2/f1 close to 1.2. Similar stimulus configurations used in the experiments have also been used for computer simulations of DPOAE in a nonlinear and active model of the cochlea. Experimental results and model simulations give strong evidence for a two-source model of DPOAE generation: The first source is the initial nonlinear interaction of the primaries close to the f2 place. The second source is caused by coherent reflection from a re-emission site at the characteristic place of the distortion product frequency. The spectral fine structure of DPOAE observed in the ear canal reflects the interaction of both these sources.

Distortion product otoacoustic emissions (DPOAE) were recorded from eight human subjects with mild to moderate cochlear hearing loss, using a frequency spacing of 48 primary pairs per octave and at a level L1 = L2 = 60 dBSPL and with a fixed ratio f2/f1. Subjects with different shapes of hearing thresholds were selected. They included subjects with near-normal hearing within only a limited frequency range, subjects with a notch in the audiogram, and subjects with a mild to moderate high-frequency loss. If the primaries were located in a region of normal or near-normal hearing, but DP frequencies were located in a region of raised thresholds, the distortion product 2 f1-f2 was still observable, but the DP fine structure disappeared. If the DP frequencies fell into a region of normal thresholds, fine structure was preserved as long as DPOAE were generated, even in cases of mild hearing loss in the region of the primaries. These experimental results give further strong evidence that, in addition to the initial source in the primary region, there is a second source at the characteristic place of fDP. Simulations in a nonlinear and active computer model for DPOAE generation indicate different generation mechanisms for the two components. The disappearance of DPOAE fine structure might serve as a more sensitive indicator of hearing impairment than the consideration of DP level alone.

This paper addresses the question of whether subjects are able to assess the perceived time-varying quality of speech material continuously. A method is introduced which is characterized by a subjective continuous rating of the perceived speech quality by moving a slider along a graphical scale. The usability of this method is illustrated with an experiment in which different sequences of sentences were degraded in quality with a Modulated Noise Reference Unit. The modulation depth was varied with time and the subject's task was to assess the perceived quality. The results indicate that subjects can monitor speech quality variations very accurately with a delay of approximately 1 s. An objective speech quality measure based on an auditory processing model was applied to predict the subjective speech quality results. The speech quality measure qC was modified to allow for time-dependent objective measurement of the speech quality. The averaged subjective response data could be modeled by the scale transformed and low-pass filtered measure qC(t) with a high degree of accuracy.

Experiments and model calculations were performed to study the influence of within-channel cues versus across-channel cues in comodulation masking release (CMR). A class of CMR experiments is considered that are characterized by a single (unmodulated or modulated) bandpass noise masker with variable bandwidth centered at the signal frequency. A modulation-filterbank model suggested by Dau et al. [J. Acoust. Soc. Am. 102, 2892-2905 (1997)] was employed to quantitatively predict the experimental data. Effects of varying masker bandwidth, center frequency, modulator bandwidth, modulator type, and signal duration on CMR were examined. In addition, the effect of band limiting the noise before or after modulation was shown to influence the CMR in the same way as a systematic variation of the modulation depth. It is demonstrated that a single-channel analysis, which analyzes only the information from one peripheral channel, quantitatively accounts for the CMR in most cases, indicating that an across-channel process is generally not necessary for simulating results from this class of CMR experiments. True across-channel processes may be found in another class of CMR experiments.

A front end for automatic speech recognizers is proposed and evaluated which is based on a quantitative model of the "effective" peripheral auditory processing. The model simulates both spectral and temporal properties of sound processing in the auditory system which were found in psychoacoustical and physiological experiments. The robustness of the auditory-based representation of speech was evaluated in speaker-independent, isolated word recognition experiments in different types of additive noise. The results show a higher robustness of the auditory front end in noise, compared to common mel-scale cepstral feature extraction. In a second set of experiments, different processing stages of the auditory front end were modified to study their contribution to robust speech signal representation in detail. The adaptive compression stage which enhances temporal changes of the input signal appeared to be the most important processing stage towards robust speech representation in noise. Low-pass filtering of the fast fluctuating envelope in each frequency band further reduces the influence of noise in the auditory-based representation of speech.

Auditory filter bandwidths and time constants were obtained with five normal-hearing subjects for different masker configurations both in the frequency and time domain for monaural and binaural listening conditions. Specifically, the masking level in the monaural condition and the interaural correlation in the binaural conditions, respectively, was changed in a sinusoidal, stepwise, and rectangular way in the frequency domain. In the corresponding experiments in the time domain, a sinusoidal and stepwise change of the masker was performed. From these results, a comparison was made across conditions to evaluate the influence of the factors "shape of transition," "monaural versus binaural," "frequency domain versus time domain," and "subject." Also, the respective data from the literature were considered using the same model assumptions and fitting strategy as used for the current data. The results indicate that the monaural auditory filter bandwidths and time constants fitted to the data are consistent across conditions both for the data included in this study and the data from the literature. No consistent relation between individual auditory filter bandwidths and time constants were found across subjects. For the binaural conditions, however, considerable differences were found in estimates of the bandwidths and time constants, respectively, across conditions. The reason for this mismatch seems to be the different detection strategies employed for the various tasks that are affected by the consistency of binaural information across frequency and time. While monaural detection performance appears to be modeled quite well with a linear filter or temporal integration window, this does not hold for the binaural conditions where both larger bandwidth and time constant estimates are found.

A multi-channel model, describing the effects of spectral and temporal integration in amplitude-modulation detection for a stochastic noise carrier, is proposed and validated. The model is based on the modulation filterbank concept which was established in the accompanying paper [Dau et al., J. Acoust. Soc. Am. 102, 2892-2905 (1997)] for modulation perception in narrow-band conditions (single-channel model). To integrate information across frequency, the detection process of the model linearly combines the channel outputs. To integrate information across time, a kind of "multiple-look" strategy, is realized within the detection stage of the model. Both data from the literature and new data are used to validate the model. The model predictions agree with the results of Eddins [J. Acoust. Soc. Am. 93, 470-479 (1993)] that the "time constants" associated with the temporal modulation transfer functions (TMTF) derived for narrow-band stimuli do not vary with carrier frequency region and that they decrease monotonically with increasing stimulus bandwidth. The model is able to predict masking patterns in the modulation-frequency domain, as observed experimentally by Houtgast [J. Acoust. Soc. Am. 85, 1676-1680 (1989)]. The model also accounts for the finding by Sheft and Yost [J. Acoust. Soc. Am. 88, 796-805 (1990)] that the long "effective" integration time constants derived from the data are two orders of magnitude larger than the time constants derived from the cutoff frequency of the TMTF. Finally, the temporal-summation properties of the model allow the prediction of data in a specific temporal paradigm used earlier by Viemeister and Wakefield [J. Acoust. Soc. Am. 90, 858-865 (1991)]. The combination of the modulation filterbank concept and the optimal decision algorithm proposed here appears to present a powerful strategy for describing modulation-detection phenomena in narrow-band and broadband conditions.

This paper presents a quantitative model for describing data from modulation-detection and modulation-masking experiments, which extends the model of the "effective" signal processing of the auditory system described in Dau et al. [J. Acoust. Soc. Am. 99, 3615-3622 (1996)]. The new element in the present model is a modulation filterbank, which exhibits two domains with different scaling. In the range 0-10 Hz, the modulation filters have a constant bandwidth of 5 Hz. Between 10 Hz and 1000 Hz a logarithmic scaling with a constant Q value of 2 was assumed. To preclude spectral effects in temporal processing, measurements and corresponding simulations were performed with stochastic narrow-band noise carriers at a high center frequency (5 kHz). For conditions in which the modulation rate (fmod) was smaller than half the bandwidth of the carrier (delta f), the model accounts for the low-pass characteristic in the threshold functions [e.g., Viemeister, J. Acoust. Soc. Am. 66, 1364-1380 (1979)]. In conditions with fmod > delta f/2, the model can account for the high-pass characteristic in the threshold function. In a further experiment, a classical masking paradigm for investigating frequency selectivity was adopted and translated to the modulation-frequency domain. Masked thresholds for sinusoidal test modulation in the presence of a competing modulation masker were measured and simulated as a function of the test modulation rate. In all cases, the model describes the experimental data to within a few dB. It is proposed that the typical low-pass characteristic of the temporal modulation transfer function observed with wide-band noise carriers is not due to "sluggishness" in the auditory system, but can instead be understood in terms of the interaction between modulation filters and the inherent fluctuations in the carrier.

A German sentence test was developed which is comprised of 20 test lists of ten sentences each. The test corpus is a selection from sentences for speech quality evaluation recorded with a male unschooled speaker. Performance-intensity curves were measured for each individual sentence in a speech-simulating babble noise with a total of 40 normal-hearing listeners. Based on these data and the phonemic transcription of the 200 sentences selected from the underlying speech corpus, 20 test lists were composed using a numerical optimization process. These 20 test lists are highly equivalent with respect to their performance-intensity curves, the number of words within each test list, the number of phonemes within each test list, and approximately the frequency distribution of the phonemes which approximates the phoneme frequency distribution of the German language. The equivalence of the respective performance-intensity curves was demonstrated in an independent experiment with 20 normal-hearing listeners. In addition, a comparison was performed between the "objective" intelligibility measurements and two "subjective" speech intelligibility rating methods employing the same materials. As a result, both subjective assessment procedures correlate highly with each other and with the "objective" procedure across sentences. This underlines the applicability and validity of the test in combination with time-saving subjective assessment methods. Moreover, the variability in performance across different sentences correlates inversely with the RMS level of the respective sentence. This indicates that an adjustment of sentence material with respect to RMS level already yield reasonably homogeneous test material with respect to intelligibility.

Narrow-band transitory evoked otoacoustic emissions (TEOAE) were recorded for nine normal hearing subjects in the presence of a broadband tone complex suppressor. Introducing a spectral notch at the frequency of the narrow-band stimulus causes the suppression effect to decrease, the more so the wider the notch. This decrease in suppression permits an estimate of the size of one critical band. One advantage of this approach is that no active participation of the subjects is required. The estimated critical bandwidth is then compared with independent estimates based on a simultaneous masking experiment, using the same stimuli. The two measures of the critical bandwidth coincide well for those six subjects with spontaneous otoacoustic emissions. However, the bandwidth estimate based on the OAE measurements is too large for the other three subjects without spontaneous emissions. Simulations of the suppression effect with a driven van der Pol oscillator with moderate undamping produce critical bandwidth estimates consistent with those observed in the psychoacoustical experiments. This allows an estimate of the "effective" amount of undamping on the basilar membrane that is required to produce the critical bandwidth observable in psychoacoustic experiments.

Speech reception thresholds (SRTs) were obtained for different azimuths of up to three interfering sound sources in a simulated ("virtual") spatial arrangement. For the SRT measurements a fast subjective threshold assessment method was used. The reduction in SRT relative to the reference threshold caused by spatial separation of target and interference signal is called the intelligibility level difference (ILD). For normal listeners, the maximum ILD was achieved at 105 degrees or 255 degrees azimuth of a single noise source when the target emanated from the front (0 degree azimuth). The ILD decreased rapidly if the number of interfering signals was increased and if they were located on opposite sides of the head. When using continuous speech as interference, this decrease of ILD was less pronounced. These findings indicate that the binaural system can suppress interference from only one azimuthal region at a time and can utilize temporal pauses in the interference to suppress a second interference from a different direction. For eight sensorineural impaired listeners, the maximum attainable ILD for one interfering noise source was reduced in a way not predictable from the audiogram. The ability to exploit pauses in the signal of an interfering talker was also reduced. Hence, hearing-impaired listeners are handicapped in realistic spatial interference situations both by an increased monaural SRT and a reduced binaural noise reduction ability.

The influence of an external tone on transitory evoked otoacoustic emissions (TEOAE) is investigated. Three different averaging techniques were used with the same acoustic stimulus paradigm. These techniques permitted the separation of those parts of the otoacoustic emission (OAE) that contribute to the transitory evoked otoacoustic emission and those parts of the OAE that are synchronized to the continuous tone. The experiments show that the total energy of the OAE is not reduced in the presence of an additional tone. The 'suppression' of TEOAEs is an effect of synchronization and the subsequent elimination of the 'suppressed' emission in the averaging procedure.

A new method for detecting the acoustic reflex that utilizes standard otoacoustic emissions recording techniques is introduced and discussed. Two successive identical tone bursts of 100 ms duration and 10 ms interstimulus interval are presented in the occluded ear canal at a repetition rate of one per second. If the acoustic reflex is elicited, the contraction of the stapedius muscle is delayed with respect to the onset of the first stimulus. Hence, the acoustic compliance in the ear canal decreases primarily during the second stimulus. The difference of the microphone signals produced by the two stimuli is computed and averaged across a certain number of repetitions of the sequence. The presentation level is increased until this difference is larger than -40 dB (with respect to the stimulus level) and if its signal-to-noise ratio exceeds 20 dB. For normal-hearing subjects, the acoustic reflex threshold measured with this method is on average 8 dB lower than in a standard clinical setup. In 5 out of the 10 tested hearing-impaired subjects, the new method could detect an acoustic reflex at one or more frequencies where no reflex was detected in the clinical setup.

Sensorneural hearing-impaired listeners suffer severely from deterioration in the processing and internal representation of acoustic signals. In order to understand this deterioration in detail, a numerical perception model was developed which is based on current functional models of the signal processing in the auditory system. To test this model, the individual's speech intelligibility in quiet and in noise was predicted. The primary input parameter of the model is the precisely measured audiogram of each listener. In a refined version of the model, additional input parameters are derived from predicting the individual's temporal forward masking and notched-noise measurements with the same model assumptions. The predictions of the perception model were compared with those of the articulation index (AI) and the speech transmission index (STI). The accuracy of prediction with the perception model is in the same range as with the AI and the STI. The model does not require a calibration function and has the advantage of a greater flexibility in including different processing deficits associated with hearing impairment. However, it requires more time for computation. For the hearing-impaired listeners examined so far the individually measured psychoacoustical parameters have only a secondary effect on the prediction as compared to the audiogram. Nevertheless, the underlying model is a first step toward a quantitative understanding of speech intelligibility and helps to distinguish between the influence of the "attenuation" and the "distortion" component of the hearing loss.

In the recent literature on dynamic compression in hearing aids, a controversial discussion about the benefit of multichannel compression with short time constants has been raised. In this paper, some experimental results are contributed to the discussion: For low signal-to-noise ratios, a detrimental effect of fast multichannel compression on speech intelligibility in normal hearing listeners is found. However, a distinct degradation in speech quality and virtually no decrease in speech intelligibility is observed for comparatively high signal-to-noise ratios and compression ratios of up to 3. Since the predictions of the Speech Transmission Index deviate substantially from these results, its applications as an objective measure of intelligibility may not be extended to compressed speech.

A novel approach for analyzing and filtering speech is described and evaluated which utilizes the "modulation spectrogram," i.e., the two-dimensional representation of modulation frequencies versus center frequency as a function of time. This approach is based on physiological findings of a tonotopical organization of modulation frequencies perpendicular to carrier frequencies as well as psychoacoustical findings of "modulation tuning curves." In addition, an interaction is assumed between the representation of modulation frequencies and the representation of auditory space as described by physiological and psychological models of binaural hearing. A noise-reduction algorithm based on this approach was implemented and tested which enhances or suppresses each combination of modulation frequency and center frequency according to its phase and intensity relation between the two input signals (i.e., both stereo channels of a dummy-head recording). When tested in several situations with interfering speakers and background noise both in anechoic and reverberant environment, the algorithm provided a small but a very robust increase in speech intelligibility which corresponds to approximately 2 dB in signal-to-noise ratio. Possible applications of this algorithm are noise reduction in adverse acoustical situations, digital hearing aids, processing schemes and preprocessing for speech recognition.

A multi-signal-processor set-up is introduced that is used for real-time implementation of digital hearing aid algorithms that operate on stereophonic (i.e., binaural) input signals and perform signal processing in the frequency domain. A multiband dynamic compression algorithm was implemented which operates in 24 critical band filter channels, allows for interaction between frequency bands and stereo channels, and is fitted to the hearing of the individual patient by a loudness scaling method. In addition, a binaural noise reduction algorithm was implemented that amplifies sound emanating from the front and suppresses lateral noise sources as well as reverberation. These algorithms were optimized with respect to their processing parameters and by minimizing the processing artifacts. Different versions of the algorithms were tested in six listeners with sensorineural hearing impairment using both subjective quality assessment methods and speech intelligibility measurements in different acoustical situations. For most subjects, linear frequency shaping was subjectively assessed to be negative, although it improved speech intelligibility in noise. Additional compression was assessed to be positive and did not deteriorate speech intelligibility as long as the processing parameters were fitted carefully. All noise reduction strategies employed here were subjectively assessed to be positive. Although the suppression of reverberation only slightly improved speech intelligibility, a combination of directional filtering and dereverberation provided a substantial improvement in speech intelligibility for most subjects and for a certain range of signal-to-noise ratios. The real-time implementation was very helpful in optimizing and testing the algorithms, and the overall results indicate that carefully designed and fitted binaural hearing aids might be very beneficial for a large number of patients.

A binaural noise-reduction algorithm for suppressing lateral noise sources was implemented in real-time and tested with normal and hearing-impaired subjects. With this algorithm, frequency bands with interaural time and amplitude differences characteristic for the "target" direction of the desired speech sound reach both ears without any attenuation. However, frequency bands with an interaural time and amplitude difference deviating from these desired values are attenuated. Speech quality assessments as well as speech intelligibility tests demonstrate a significant noise reduction of up to about 5 dB in signal-to-noise ratio for a dummy-head recorded "cocktail-party situation" in an anechoic chamber. For two additional acoustical situations, however, the subjective speech quality from the algorithm degrades with increasing reverberation, while the output from a real-time "dereverberation" algorithm was ranked higher with increasing reverberation. The results with hearing-impaired listeners suggest a combination of both algorithms, while stressing the importance of binaural listening and binaural hearing aids for the rehabilitation of impaired listeners in noisy environments.

The work here presents the first part of a prospective study regarding the clinical use of evoked otoacoustic emissions (EOAE) in adults. Sixty subjects with normal hearing and 160 patients suffering from cochlear hearing loss were tested. The results were used to develop and optimize analysis criteria for the emissions, based on their physical properties. A short-time Fourier analysis was performed so that the EOAE intensity in time and frequency domains could be observed simultaneously. A comparison of this data with the individual thresholds of hearing showed the importance of the EOAE level. However, the bandwidth of the EOAE demonstrated an even steeper transition between normal and hearing-impaired subjects. These findings suggest that this bandwidth is a better criterion for the detection of an EOAE. On the other hand, both parameters correlated weakly with the hearing threshold and the differences between subjects were very large. A prediction of hearing loss based on EOAE results is impossible with the wideband click stimulation used here.

The shape and the effective bandwidth of the auditory filter at 500 Hz was examined for binaural and monaural tone-in-noise detection experiments in four normal listeners. In the binaural condition, a broadband noise with an interaural phase difference of 0 below and an interaural phase difference of pi above a certain "edge frequency" was employed to mask a 500-Hz probe tone with an interaural phase pi (denoted as No pi S pi). The threshold of the probe tone as a function of the edge frequency in this configuration and in a configuration with an inverted interaural phase of the masker (denoted as N pi oS pi) was fitted by assuming different filter shapes and optimizing their respective parameters. In an analogous monaural experiment, the spectral power density of the masker was 15 dB lower below the "edge frequency" or 15 dB lower above this frequency, respectively. Several filter characteristics with two free parameters describe the data almost equally well. Their equivalent rectangular bandwidths (ERB) show considerably more variations between filter shapes than the 10-dB bandwidth and the 90% bandwidth values (i.e., the bandwidths encompassing 90% of the integrated area above and below the center frequency). This indicates that either of these two bandwidth parameters is more appropriate for comparing auditory filter bandwidths than the ERB. For the rounded exponential filter, the 90% bandwidth averages to 147 Hz in the binaural and to 125 Hz in the monaural condition. These values are up to 12% higher if off-frequency detection is accounted for. Our general finding of auditory filter bandwidths in the binaural conditions exceeding the monaural bandwidths by approximately 20% may be caused by two factors: First, off-frequency detection may be performed in monaural, but not in binaural detections tasks and second, the random interaural mismatch in binaural noise reduction processes fluctuates slowly and thus modulates and spectrally smears the output signal of the binaural noise reduction process.

The purpose of this study was the evaluation of evoked otoacoustic emissions (EOAE) for the screening of infant hearing. EOAE and brainstem-evoked response audiometry (BERA) were performed in 40 infants at a special care unit at the Universitäts-Klinik, Göttingen. Both examinations were performed under non-optimal conditions, reflecting a realistic screening situation. The purpose was to determine robust analysis criteria and possible error sources. A comparison of both tests shows that sensitivity and specificity are insufficient for reliable screening. The EOAE test does not yield the threshold of hearing. There is no sharp boundary with the BERA threshold for the detectability of an EOAE. Sensitivity and specificity have not been accurately determined so far due to the relatively small number of infants with impaired hearing tested. Nevertheless, the results show that EOAE should be applied as a supplement to conventional audiometry for infants. It is capable of excluding conductive and severe cochlear hearing losses and is by itself not a reliable screening test.

The threshold of a short interaurally phase-inverted probe tone (20 ms, 500 Hz, S pi) was obtained in the presence of a 750-ms noise masker that was switched after 375 ms from interaurally phase-inverted (N pi) to interaurally in-phase (No). As the delay between probe-tone offset and noise phase transition is increased, the threshold decays from the N pi S pi threshold (masking level difference = 0 dB) to the No S pi threshold (masking level difference = 15 dB). The decay in this "binaural" situation is substantially slower than in a comparable "monaural" situation, where the interaural phase of the masker is held constant (N pi), but the level of the masker is reduced by 15 dB. The prolonged decay provides evidence for additional binaural sluggishness associated with "binaural forward masking." In a second experiment, "binaural backward masking" is studied by time reversing the maskers described above. Again, the situation where the phase is switched from No to N pi exhibits a slower transition than the situation with constant interaural phase (N pi) and a 15-dB increase in the level of the masker. The data for the binaural situations are compatible with the results of a related experiment, previously reported by Grantham and Wightman [J. Acoust. Soc. Am. 65, 1509-1517 (1979)] and are well fit by a model that incorporates a double-sided exponential temporal integration window.

Data from a simple tone-in-noise simultaneous masking task were used to evaluate each of two common adaptive staircase rules (a "1 up 2 down" rule and a "1 up 3 down" rule) and the parameter estimation by sequential testing (PEST) technique in combination with each of two psychophysical procedures [a two-alternative forced-choice (2AFC) and a three-alternative forced-choice (3AFC) procedure]. These human data were compared to predictions generated by a mathematical model based on Markov theory. The model predicts that threshold estimates obtained with the adaptive techniques should be equal to those derived with equivalent "fixed signal level" techniques. However, the human data indicate that the adaptive techniques tend to yield lower thresholds. The model predicts that the standard error of a threshold estimate obtained from an adaptive technique will decrease and approach zero as the number of trials used to compute the estimate increases. The human data show greater variability than predicted and approach a nonzero value as the number of trials increases. The predictions of the model suggest that the commonly used combination of the 2AFC procedure and the 1 up 2 down rule is the least efficient method of estimating a threshold and that the 3AFC procedure in combination with the 1 up 3 down rule is the most efficient method. The human data are less consistent, but generally show the combination of the 2AFC procedure and the 1 up 2 down rule to be one of the least efficient methods. Possible explanations for the differences between the model's predictions and the human data, as well as suggestions for laboratory practice, are discussed.