Effect of auditory efferent time-constant duration on speech recognition in noise.

The human auditory efferent system may play a role in improving speech-in-noise recognition with an associated range of time constants. Computational auditory models with efferent-inspired feedback demonstrate improved speech-in-noise recognition with long efferent time constants (2000 ms). This study used a similar model plus an Automatic Speech Recognition (ASR) system to investigate the role of shorter time constants. ASR speech recognition in noise improved with efferent feedback (compared to no-efferent feedback) for both short and long efferent time constants. For some signal-to-noise ratios, speech recognition in noise improved as efferent time constants were increased from 118 to 2000 ms.

Exploration of a physiologically-inspired hearing-aid algorithm using a computer model mimicking impaired hearing.

OBJECTIVE: To use a computer model of impaired hearing to explore the effects of a physiologically-inspired hearing-aid algorithm on a range of psychoacoustic measures. DESIGN: A computer model of a hypothetical impaired listener's hearing was constructed by adjusting parameters of a computer model of normal hearing. Absolute thresholds, estimates of compression, and frequency selectivity (summarized to a hearing profile) were assessed using this model with and without pre-processing the stimuli by a hearing-aid algorithm. The influence of different settings of the algorithm on the impaired profile was investigated. To validate the model predictions, the effect of the algorithm on hearing profiles of human impaired listeners was measured. STUDY SAMPLE: A computer model simulating impaired hearing (total absence of basilar membrane compression) was used, and three hearing-impaired listeners participated. RESULTS: The hearing profiles of the model and the listeners showed substantial changes when the test stimuli were pre-processed by the hearing-aid algorithm. These changes consisted of lower absolute thresholds, steeper temporal masking curves, and sharper psychophysical tuning curves. CONCLUSION: The hearing-aid algorithm affected the impaired hearing profile of the model to approximate a normal hearing profile. Qualitatively similar results were found with the impaired listeners' hearing profiles.

Hearing dummies: individualized computer models of hearing impairment.

UNLABELLED: Objective: Our aim was to explore the usage of individualized computer models to simulate hearing loss based on detailed psychophysical assessment and to offer hypothetical diagnoses of the underlying pathology. DESIGN: Individualized computer models of normal and impaired hearing were constructed and evaluated using the psychophysical data obtained from human listeners. Computer models of impaired hearing were generated to reflect the hypothesized underlying pathology (e.g. dead regions, outer hair cell dysfunction, or reductions in endocochlear potential). These models were evaluated in terms of their ability to replicate the original patient data. STUDY SAMPLE: Auditory profiles were measured for two normal and five hearing-impaired listeners using a battery of three psychophysical tests (absolute thresholds, frequency selectivity, and compression). RESULTS: The individualized computer models were found to match the data. Useful fits to the impaired profiles could be obtained by changing only a single parameter in the model of normal hearing. Sometimes, however, it was necessary to include an additional dead region. CONCLUSION: The creation of individualized computer models of hearing loss can be used to simulate auditory profiles of impaired listeners and suggest hypotheses concerning the underlying peripheral pathology.

The robustness of speech representations obtained from simulated auditory nerve fibers under different noise conditions.

Different methods of extracting speech features from an auditory model were systematically investigated in terms of their robustness to different noises. The methods either computed the average firing rate within frequency channels (spectral features) or inter-spike-intervals (timing features) from the simulated auditory nerve response. When used as the front-end for an automatic speech recognizer, timing features outperformed spectral features in Gaussian noise. However, this advantage was lost in babble, because timing features extracted the spectro-temporal structure of babble noise, which is similar to the target speaker. This suggests that different feature extraction methods are optimal depending on the background noise.

Acquisition of auditory profiles for good and impaired hearing.

OBJECTIVE: The aim of this study was to develop a user-friendly way of measuring patients' threshold and supra-threshold hearing, with potential for application in clinical research. The end-product of these tests is a graphical profile summarizing absolute threshold, frequency selectivity, and compression characteristics across a spectrum of frequencies (0.25-6 kHz). DESIGN: A battery of three psychophysical hearing tests consisted of measures of absolute threshold, frequency selectivity, and compression. An automated, cued, single-interval, adaptive tracking procedure was employed. The tests results were collated and used to generate a readily visualized 'profile' for each listener. STUDY SAMPLE: Participants were 83 adults (57 impaired-hearing and 26 good-hearing, age 20-75 years). RESULTS: Listeners tolerated the tests well. Single-ear profiles were obtained in an average of 74 minutes testing time (range 46-120 minutes). The variability of individual measurements was low. Substantial differences between normal and impaired listeners and also among the impaired listeners were observed. Qualitative differences in compression and frequency-selectivity were seen that could not be predicted by threshold measurements alone. CONCLUSIONS: The hearing profiles are informative with respect to supra-threshold hearing performance and the information is easily accessible through the graphical display. Further development is required for routine use in a clinical context.

A frequency-selective feedback model of auditory efferent suppression and its implications for the recognition of speech in noise.

The potential contribution of the peripheral auditory efferent system to our understanding of speech in a background of competing noise was studied using a computer model of the auditory periphery and assessed using an automatic speech recognition system. A previous study had shown that a fixed efferent attenuation applied to all channels of a multi-channel model could improve the recognition of connected digit triplets in noise [G. J. Brown, R. T. Ferry, and R. Meddis, J. Acoust. Soc. Am. 127, 943-954 (2010)]. In the current study an anatomically justified feedback loop was used to automatically regulate separate attenuation values for each auditory channel. This arrangement resulted in a further enhancement of speech recognition over fixed-attenuation conditions. Comparisons between multi-talker babble and pink noise interference conditions suggest that the benefit originates from the model's ability to modify the amount of suppression in each channel separately according to the spectral shape of the interfering sounds.

The psychophysics of absolute threshold and signal duration: a probabilistic approach.

The absolute threshold for a tone depends on its duration; longer tones have lower thresholds. This effect has traditionally been explained in terms of "temporal integration" involving the summation of energy or perceptual information over time. An alternative probabilistic explanation of the process is formulated in terms of simple equations that predict not only the time ∕ duration dependence but also the shape of the psychometric function at absolute threshold. It also predicts a tight relationship between these two functions. Measurements made using listeners with either normal or impaired hearing show that the probabilistic equations adequately fit observed threshold-duration functions and psychometric functions. The mathematical formulation implies that absolute threshold can be construed as a two-valued function: (a) gain and (b) sensory threshold, and both parameters can be estimated from threshold-duration data. Sensorineural hearing impairment is sometimes associated with a smaller threshold ∕ duration effect and sometimes with steeper psychometric functions. The equations explain why these two effects are expected to be linked. The probabilistic approach has the potential to discriminate between hearing deficits involving gain reduction and those resulting from a raised sensory threshold.

A computer model of auditory efferent suppression: implications for the recognition of speech in noise.

The neural mechanisms underlying the ability of human listeners to recognize speech in the presence of background noise are still imperfectly understood. However, there is mounting evidence that the medial olivocochlear system plays an important role, via efferents that exert a suppressive effect on the response of the basilar membrane. The current paper presents a computer modeling study that investigates the possible role of this activity on speech intelligibility in noise. A model of auditory efferent processing [Ferry, R. T., and Meddis, R. (2007). J. Acoust. Soc. Am. 122, 3519-3526] is used to provide acoustic features for a statistical automatic speech recognition system, thus allowing the effects of efferent activity on speech intelligibility to be quantified. Performance of the "basic" model (without efferent activity) on a connected digit recognition task is good when the speech is uncorrupted by noise but falls when noise is present. However, recognition performance is much improved when efferent activity is applied. Furthermore, optimal performance is obtained when the amount of efferent activity is proportional to the noise level. The results obtained are consistent with the suggestion that efferent suppression causes a "release from adaptation" in the auditory-nerve response to noisy speech, which enhances its intelligibility.

A simple single-interval adaptive procedure for estimating thresholds in normal and impaired listeners.

This report presents a single-interval adaptive procedure for measuring thresholds in untrained normal and impaired listeners. The accuracy of the procedure is evaluated using Monte Carlo methods and human data allowing a method to be proposed for deciding in advance the number of trials required to achieve a specified level of accuracy. The number of trials depends on the slope of the psychometric function. The slope of the psychometric function is evaluated in normal and impaired listeners, and is found to give a useful guide to the required number of trials. The single-interval up/down procedure is subsequently compared with two other popular traditional methods (two-interval forced-choice, two-down/one-up and maximum-likelihood procedures) and is shown to yield similar thresholds and be more efficient.

A cascade autocorrelation model of pitch perception.

Autocorrelation algorithms, in combination with computational models of the auditory periphery, have been successfully used to predict the pitch of a wide range of complex stimuli. However, new stimuli are frequently offered as counterexamples to the viability of this approach. This study addresses the issue of whether in the light of these challenges the predictive power of autocorrelation can be preserved by changes to the peripheral model and the computational algorithm. An existing model is extended by the addition of a low-pass filter of the summary integration of the individual within-channel autocorrelations. Other recent developments are also incorporated, including nonlinear processing on the basilar membrane and the use of integration time constants that are proportional to the autocorrelation lags. The modified and extended model predicts with reasonable success the pitches of a range of stimuli that have proved problematic for earlier implementations of the autocorrelation principle. The evaluation stimuli include short tone sequences, click trains consisting of alternating interclick intervals, click trains consisting of mixtures of regular and irregular intervals, shuffled click trains, and transposed tones.

Adversarial relationship between combined medial olivocochlear (MOC) and middle-ear-muscle (MEM) reflexes and alarm-in-noise detection thresholds under negative signal-to-noise ratios (SNRs).

The role of auditory efferent feedback from the medial olivocochlear system (MOCS) and the middle-ear-muscle (MEM) reflex in tonal detection tasks for humans in the presence of noise is not clearly understood. Past studies have yielded inconsistent results on the relationship between efferent feedback and tonal detection thresholds. This study attempts to address this inconsistency. Fifteen human subjects with normal hearing participated in an experiment where they were asked to identify an alarm signal in the presence of 80 dBA background (pink) noise. Masked detection thresholds were estimated using the method of two-interval forced choice (2IFC). Contralateral suppression of transient-evoked otoacoustic emissions (TEOAEs) was measured to estimate the strength of auditory efferent feedback. Subsequent correlation analysis revealed that the contralateral suppression of TEOAEs was significantly negatively correlated (r = -0.526, n = 15, p = 0.0438) with alarm-in-noise (AIN) detection thresholds under negative signal-to-noise conditions. The result implies that the stronger the auditory efferent feedback, the worse the detection thresholds and thus the poorer the tonal detection performance in the presence of loud noise.

A computer model of medial efferent suppression in the mammalian auditory system.

Stimulation of the olivocochlear bundle reduces basilar membrane displacement, driven auditory nerve activity, and compound action potential (CAP) response to acoustic stimulation. These effects were simulated using a computer model of the auditory periphery. The model simulates the medial efferent activity by attenuating the basilar membrane response. The model was evaluated against three animal studies reporting measurements at three levels of the auditory system; basilar membrane, single auditory nerve fibers and whole auditory nerve CAP. The CAP data included conditions where tones were masked by noise and "unmasked" by stimulation of the olivocochlear bundle. The model was able to simulate the data both qualitatively and quantitatively. As a consequence, it may be a suitable platform for studying the contribution of the efferent system to auditory processing of more complex auditory sounds in distracting backgrounds.

Further studies on the dual-resonance nonlinear filter model of cochlear frequency selectivity: responses to tones.

A number of phenomenological models that simulate the response of the basilar membrane motion can reproduce a range of complex features observed in animal measurements over different sites along its cochlea. The present report shows a detailed analysis of the responses to tones of an improved model based on a dual-resonance nonlinear filter. The improvement consists in adding a third path formed by a linear gain and an all-pass filter. This improvement allows the model to reproduce the gain and phase plateaus observed empirically at frequencies above the best frequency. The middle ear was simulated by using a digital filter based on the empirical impulse response of the chinchilla stapes. The improved algorithm is evaluated against observations of basilar membrane responses to tones at seven different sites along the chinchilla cochlear partition. This is the first time that a whole set of animal observations using the same technique has been available in one species for modeling. The resulting model was able to simulate amplitude and phase responses to tones from basal to apical sites. Linear regression across the optimized parameters for seven different sites was used to generate a complete filterbank.

Cochlear compression in listeners with moderate sensorineural hearing loss.

Psychophysical estimates of basilar membrane (BM) responses suggest that normal-hearing (NH) listeners exhibit constant compression for tones at the characteristic frequency (CF) across the CF range from 250 to 8000 Hz. The frequency region over which compression occurs is broadest for low CFs. This study investigates the extent that these results differ for three hearing-impaired (HI) listeners with sensorineural hearing loss. Temporal masking curves (TMCs) were measured over a wide range of probe (500-8000 Hz) and masker frequencies (0.5-1.2 times the probe frequency). From these, estimated BM response functions were derived and compared with corresponding functions for NH listeners. Compressive responses for tones both at and below CF occur for the three HI ears across the CF range tested. The maximum amount of compression was uncorrelated with absolute threshold. It was close to normal for two of the three HI ears, but was either slightly (at CFs < or =1000 Hz) or considerably (at CFs > or =4000 Hz) reduced for the third ear. Results are interpreted in terms of the relative damage to inner and outer hair cells affecting each of the HI ears. Alternative interpretations for the results are also discussed, some of which cast doubts on the assumptions of the TMC-based method and other behavioral methods for estimating human BM compression.

Tinnitus and patterns of hearing loss.

Tinnitus is strongly linked with the presence of damaged hearing. However, it is not known why tinnitus afflicts only some, and not all, hearing-impaired listeners. One possibility is that tinnitus patients have specific inner ear damage that triggers tinnitus. In this study, differences in cochlear function inferred from psychophysical measures were measured between hearing-impaired listeners with tinnitus and hearing-impaired listeners without tinnitus. Despite having similar average hearing loss, tinnitus patients were observed to have better frequency selectivity and compression than those without tinnitus. The results suggest that the presence of subjective tinnitus may not be strongly associated to outer hair cell impairment, at least where hearing impairment is evident. The results also show a different average pattern of hearing impairment amongst the tinnitus patients, consistent with the suggestion that inner hair cell dysfunction with subsequent reduced auditory innervation is a possible trigger of tinnitus.

Off-frequency listening in subjects with chronic tinnitus.

The occurrence of subjective tinnitus has been linked to cochlear damage, as most tinnitus patients have impaired hearing, and animal studies have shown that the induction of hearing loss can lead to behavioural signs of tinnitus. In tinnitus patients, the pure-tone audiogram is the main source of information about cochlear damage, but hearing thresholds alone may not adequately reflect its magnitude. Etchelecou et al. (2011) reported that the majority of patients with acute tinnitus post impulse noise exposure showed off-frequency listening (OFL), which is not readily observed in pure-tone audiograms. We investigated the possibility of OFL occurring in subjects with chronic tinnitus by testing twenty subjects who had experienced tinnitus for more than a year. OFL was assessed by measuring psychophysical tuning curves using a forward-masking paradigm. OFL occurred in 13 out of 20 subjects, 12 of whom also did not perceive frequencies above 8 kHz. Such unresponsive frequencies (UFs) were also present in three subjects without OFL. The tinnitus spectrum generally reached its highest values at the edge of or within the frequency regions with OFL or UFs, but there was no significant correlation between edge frequencies and the frequency with the highest tinnitus pitch similarity rating. When OFL and UFs were taken as evidence for cochlear dead regions, 16/20 subjects passed the criterion for cochlear dead regions. The remaining four subjects showed neither OFL nor UFs.

The role of auditory nerve innervation and dendritic filtering in shaping onset responses in the ventral cochlear nucleus.

Neurons in the ventral cochlear nucleus (VCN) that respond primarily at the onset of a pure tone stimulus show diversity in terms of peri-stimulus-time-histograms (PSTHs), rate-level functions, frequency tuning, and also their responses to broad band noise. A number of different mechanisms have been proposed as contributing to the onset characteristic: e.g. coincidence, depolarisation block, and low-threshold potassium currents. We show that a simple point neuron receiving convergent inputs from high-spontaneous rate auditory nerve (AN) fibers, with no special currents and no peri-stimulatory shifts in firing threshold, is sufficient to produce much of the diversity seen experimentally. Three sub-classes of onset PSTHs: onset-ideal (OI), onset-chopper (OC) and onset-locker (OL) are reproduced by variations in innervation patterns and dendritic filtering. The factors shaping responses were explored by systematically varying key parameters. An OI response in this model requires a narrow range of AN input best frequencies (BF) which only produce supra-threshold depolarizations during the stimulus onset. For OC and OL responses, receptive fields were wider. Considerable low pass filtering of AN inputs away from BF results in an OL, whilst relatively unfiltered inputs produce an OC response. Rate-level functions in response to pure tones can be sloping, or plateau. These can be also reproduced in the model by the manipulation of the AN innervation. The model supports the coincidence detection hypothesis, and suggests that differences in excitatory innervation and dendritic filtering constant are important factors to consider when accounting for the variation in response characteristics seen in VCN onset units.

Reply to comment on "Auditory-nerve first-spike latency and auditory absolute threshold: a computer model".

Krisha [J. Acoust. Soc. Am., in press (2006)] has commented that an explanation based on presynaptic calcium accumulation at the inner hair cell is an incorrect explanation for the success of a model of the auditory periphery [Meddis, R., J. Acoustic. Soc. Am. 119, 406-417 (2006)] in explaining data on first-spike auditory nerve latency. This reply accepts the criticism and accepts the strength of an alternative explanation based on expected latencies in random sequences of low-probability events. This reply also goes on briefly to explore the application of this argument to other phenomena, including the dependence of absolute auditory threshold on the duration of the stimulus. This has wide-ranging implications for the concept of "temporal integration" in psychophysics.

Auditory-nerve first-spike latency and auditory absolute threshold: a computer model.

A computer model of the auditory periphery was used to address the question of what constitutes the physiological substrate of absolute auditory threshold. The model was first evaluated to show that it is consistent with experimental findings that auditory-nerve fiber spikes can be predicted to occur when the running integral of stimulus pressure reaches some critical value [P. Heil and H. Neubauer, J. Neurosci. 15, 7404-7415 (2001)]. It was then modified to examine two ways in which the accumulation and clearance of receptor presynaptic calcium might explain this effect. Both methods gave results that matched the animal data. It was also shown how the rate of clearance of presynaptic calcium could be used to explain the origin of differences between low and high spontaneous-rate fiber types. When spiking activity is aggregated across a number of similar high spontaneous-rate fibers and used as the input to a model of a cochlear nucleus coincidence neuron, its response can be used to judge whether or not a stimulus is present. A simulated psychophysical experiment then demonstrated that this simple decision procedure can reproduce measurements of absolute auditory threshold for tones in quiet where the threshold is a joint function of both time and level.

Virtual pitch in a computational physiological model.

A computational model of nervous activity in the auditory nerve, cochlear nucleus, and inferior colliculus is presented and evaluated in terms of its ability to simulate psychophysically-measured pitch perception. The model has a similar architecture to previous autocorrelation models except that the mathematical operations of autocorrelation are replaced by the combined action of thousands of physiologically plausible neuronal components. The evaluation employs pitch stimuli including complex tones with a missing fundamental frequency, tones with alternating phase, inharmonic tones with equally spaced frequencies and iterated rippled noise. Particular attention is paid to differences in response to resolved and unresolved component harmonics. The results indicate that the model is able to simulate qualitatively the related pitch-perceptions. This physiological model is similar in many respects to autocorrelation models of pitch and the success of the evaluations suggests that autocorrelation models may, after all, be physiologically plausible.