Categorical loudness scaling and equal-loudness contours in listeners with normal hearing and hearing loss

Toward parametric Bayesian adaptive procedures for multi-frequency categorical loudness scaling

A series of Bayesian adaptive procedures to estimate loudness growth across a wide frequency range from individual listeners was developed, and these procedures were compared. Simulation experiments were conducted based on multinomial psychometric functions for categorical loudness scaling across ten test frequencies estimated from 61 listeners with normal hearing and 87 listeners with sensorineural hearing loss. Adaptive procedures that optimized the stimulus selection based on the interim estimates of two types of category-boundary models were tested. The first type of model was a phenomenological model of category boundaries adopted from previous research studies, while the other type was a data-driven model derived from a previously collected set of categorical loudness scaling data. An adaptive procedure without Bayesian active learning was also implemented. Results showed that all adaptive procedures provided convergent estimates of the loudness category boundaries and equal-loudness contours between 250 and 8000 Hz. Performing post hoc model fitting, using the data-driven model, on the collected data led to satisfactory accuracies, such that all adaptive procedures tested in the current study, independent of modeling approach and stimulus-selection rules, were able to provide estimates of the equal-loudness-level contours between 20 and 100 phons with root-mean-square errors typically under 6 dB after 100 trials.

Objective evaluation of intracochlear electrocochleography: repeatability, thresholds, and tonotopic patterns

Introduction

Intracochlear electrocochleography (ECochG) is increasingly being used to measure residual inner ear function in cochlear implant (CI) recipients. ECochG signals reflect the state of the inner ear and can be measured during implantation and post-operatively. The aim of our study was to apply an objective deep learning (DL)-based algorithm to assess the reproducibility of longitudinally recorded ECochG signals, compare them with audiometric hearing thresholds, and identify signal patterns and tonotopic behavior.

Methods

We used a previously published objective DL-based algorithm to evaluate post-operative intracochlear ECochG signals collected from 21 ears. The same measurement protocol was repeated three times over 3 months. Additionally, we measured the pure-tone thresholds and subjective loudness estimates for correlation with the objectively detected ECochG signals. Recordings were made on at least four electrodes at three intensity levels. We extracted the electrode positions from computed tomography (CT) scans and used this information to evaluate the tonotopic characteristics of the ECochG responses.

Results

The objectively detected ECochG signals exhibited substantial repeatability over a 3-month period (bias-adjusted kappa, 0.68; accuracy 83.8%). Additionally, we observed a moderate-to-strong dependence of the ECochG thresholds on audiometric and subjective hearing levels. Using radiographically determined tonotopic measurement positions, we observed a tendency for tonotopic allocation with a large variance. Furthermore, maximum ECochG amplitudes exhibited a substantial basal shift. Regarding maximal amplitude patterns, most subjects exhibited a flat pattern with amplitudes evenly distributed over the electrode carrier. At higher stimulation frequencies, we observed a shift in the maximum amplitudes toward the basal turn of the cochlea.

Conclusions

We successfully implemented an objective DL-based algorithm for evaluating post-operative intracochlear ECochG recordings. We can only evaluate and compare ECochG recordings systematically and independently from experts with an objective analysis. Our results help to identify signal patterns and create a better understanding of the inner ear function with the electrode in place. In the next step, the algorithm can be applied to intra-operative measurements.

An intracochlear electrocochleography dataset - from raw data to objective analysis using deep learning

Electrocochleography (ECochG) measures electrophysiological inner ear potentials in response to acoustic stimulation. These potentials reflect the state of the inner ear and provide important information about its residual function. For cochlear implant (CI) recipients, we can measure ECochG signals directly within the cochlea using the implant electrode. We are able to perform these recordings during and at any point after implantation. However, the analysis and interpretation of ECochG signals are not trivial. To assist the scientific community, we provide our intracochlear ECochG data set, which consists of 4,924 signals recorded from 46 ears with a cochlear implant. We collected data either immediately after electrode insertion or postoperatively in subjects with residual acoustic hearing. This data descriptor aims to provide the research community access to our comprehensive electrophysiological data set and algorithms. It includes all steps from raw data acquisition to signal processing and objective analysis using Deep Learning. In addition, we collected subject demographic data, hearing thresholds, subjective loudness levels, impedance telemetry, radiographic findings, and classification of ECochG signals.

Evaluation of Remote Categorical Loudness Scaling

PURPOSE

The aims of this study were to (a) demonstrate the feasibility of administering categorical loudness scaling (CLS) tests in a remote setting, (b) assess the reliability of remote compared with laboratory CLS results, and (c) provide preliminary evidence of the validity of remote CLS testing.

METHOD

CLS data from 21 adult participants collected in a home setting were compared to CLS data collected in a laboratory setting from previous studies. Five participants took part in studies in both settings. Precalibrated equipment was delivered to participants who performed headphone output level checks and measured ambient noise levels. After a practice run, CLS measurements were collected for two runs at 1 and 4 kHz.

RESULTS

Mean headphone output levels were within 1.5 dB of the target calibration level. Mean ambient noise levels were below the target level. Within-run variability was similar between the two settings, but across-run bias was smaller for data collected in the laboratory setting compared with the remote setting. Systematic differences in CLS functions were not observed for the five individuals who participated in both settings.

CONCLUSIONS

This study demonstrated that precise stimulus levels can be delivered and background noise levels can be controlled in a home environment. Across-run bias for remote CLS was larger than for in-laboratory CLS, indicating that further work is needed to improve the reliability of CLS data collected in remote settings. Supplemental Material https://doi.org/10.23641/asha.17131856.

Weakened Cochlear Nonlinearity During Human Aging and Perceptual Correlates

OBJECTIVE

As humans age, compressive nonlinearity-a hallmark of healthy cochlear function-changes. The nonlinear distortion-component of the distortion product otoacoustic emission (DPOAE) provides a noninvasive gauge of cochlear nonlinearity. Earlier published work has suggested that weakened nonlinearity begins in middle age; the current work extends this investigation into the eight decade of life using advanced DPOAE data collection and analysis methods as well as multiple metrics of nonlinearity, including a test of loudness scaling.

DESIGN

The 2f1-f2 DPOAE was recorded in 20 young adults, 25 middle-aged adults and 32 older adults from f2 = 0.78 to 9.4 kHz with primary tones (f2/f1 = 1.22) swept upward at a rate of 0.5 octave/sec. Only frequencies with audiometric thresholds ≤20 dB HL were included in the analysis and to the extent possible, ears were audiometrically matched to eliminate hearing threshold as a contributing factor to the observed age effects. Input/output functions were generated for the separated distortion-component of the DPOAE to probe compressive nonlinearity of the cochlea, and ipsilateral suppression of the DPOAE was conducted to probe two-tone suppression. To investigate the perceptual effects of weakening nonlinearity on loudness perception, the same subjects performed categorical loudness scaling. Age effects on both DPOAE and loudness scaling variables were assessed, and correlations were conducted between key OAE and perceptual metrics.

RESULTS

Age × Frequency ANOVAs revealed that the compression knee of the DPOAE I/O function occurred at higher stimulus levels in both groups of older adults compared to young adults, suggesting an expanded linear range with aging; also, the compressive slope (growth beyond the knee point) was steeper in older-adults compared to young adults. These results were most notable at high frequencies. ANOVAs including age and auditory threshold as factors confirmed that the age effect observed was independent of threshold. Additionally, in smaller subsets of subjects with audiometrically matched data, these same trends persisted, further ruling out hearing threshold as an influential factor. The growth of DPOAE ipsilateral suppression was shallower near 4 kHz in middle-aged and older adults compared to young adults and elevated suppression thresholds were observed. Results of categorical loudness scaling showed steeper growth of loudness for older adults and, at fixed sensation levels (dB SL), the older-adult group rated tones as louder than did their young-adult counterparts, suggesting abnormal loudness growth and perception. Several correlations between the compression knee of the DPOAE I/O function and key metrics of loudness scaling were significant and accounted for up to one-third of the variance.

CONCLUSIONS

Results indicate that the aging cochlea begins to show weakened nonlinearity in middle age and it progressively weakens further into senescence. The perceptual impact of weakened nonlinearity during aging is manifested as abnormal loudness judgments; that is, in older-adult ears, a tone considered comfortable or medium in young-adult ears can be considered loud. The biophysical origin of this weakened nonlinearity is not known. It is hypothesized to reflect aging-related damage to, or loss of, outer hair cells and their stereocilia. More work is warranted to better define the perceptual impact of a linearized cochlear response in older adults and to consider how this deficit might impact the fitting of hearing aids and other intervention strategies.

Maximum Expected Information Approach for Improving Efficiency of Categorical Loudness Scaling

Categorical loudness scaling (CLS) measures provide useful information about an individual's loudness perception across the dynamic range of hearing. A probability model of CLS categories has previously been described as a multi-category psychometric function (MCPF). In the study, a representative "catalog" of potential listener MCPFs was used in conjunction with maximum-likelihood estimation to derive CLS functions for participants with normal hearing and with hearing loss. The approach of estimating MCPFs for each listener has the potential to improve the accuracy of the CLS measurements, particularly when a relatively low number of data points are available. The present study extends the MCPF approach by using Bayesian inference to select stimulus parameters that are predicted to yield maximum expected information (MEI) during data collection. The accuracy and reliability of the MCPF-MEI approach were compared to the standardized CLS measurement procedure (ISO 16832:2006, 2006). A non-adaptive, fixed-level, paradigm served as a "gold-standard" for this comparison. The test time required to obtain measurements in the standard procedure is a major barrier to its clinical uptake. Test time was reduced from approximately 15 min to approximately 3 min with the MEI-adaptive procedure. Results indicated that the test-retest reliability and accuracy of the MCPF-MEI adaptive procedures were similar to the standardized CLS procedure. Computer simulations suggest that the reliability and accuracy of the MEI procedure were limited by intrinsic uncertainty of the listeners represented in the MCPF catalog. In other words, the MCPF provided insufficient predictive power to significantly improve adaptive-tracking efficiency under practical conditions. Concurrent optimization of both the MCPF catalog and the MEI-adaptive procedure have the potential to produce better results. Regardless of the adaptive-tracking method used in the CLS procedure, the MCPF catalog remains clinically useful for enabling maximum-likelihood determination of loudness categories.

Potential Consequences of Spectral and Binaural Loudness Summation for Bilateral Hearing Aid Fitting

Aversiveness of loud sounds is a frequent complaint by hearing aid users, especially when fitted bilaterally. This study investigates whether loudness summation can be held responsible for this finding. Two aspects of loudness summation should be taken into account: spectral loudness summation for broadband signals and binaural loudness summation for signals that are presented binaurally. In this study, the effect of different symmetrical hearing losses was studied. Measurements were obtained with the widely used technique of Adaptive Categorical Loudness Scaling. For large bandwidths, spectral loudness summation for hearing-impaired listeners was found to be greater than that for normal-hearing listeners, both for monaurally and binaurally presented signals. For binaural loudness summation, the effect of hearing loss was not significant. In all cases, individual differences were substantial.

Using Thresholds in Noise to Identify Hidden Hearing Loss in Humans

OBJECTIVES

Recent animal studies suggest that noise-induced synaptopathy may underlie a phenomenon that has been labeled hidden hearing loss (HHL). Noise exposure preferentially damages low spontaneous-rate auditory nerve fibers, which are involved in the processing of moderate- to high-level sounds and are more resistant to masking by background noise. Therefore, the effect of synaptopathy may be more evident in suprathreshold measures of auditory function, especially in the presence of background noise. The purpose of this study was to develop a statistical model for estimating HHL in humans using thresholds in noise as the outcome variable and measures that reflect the integrity of sites along the auditory pathway as explanatory variables. Our working hypothesis is that HHL is evident in the portion of the variance observed in thresholds in noise that is not dependent on thresholds in quiet, because this residual variance retains statistical dependence on other measures of suprathreshold function.

DESIGN

Study participants included 13 adults with normal hearing (≤15 dB HL) and 20 adults with normal hearing at 1 kHz and sensorineural hearing loss at 4 kHz (>15 dB HL). Thresholds in noise were measured, and the residual of the correlation between thresholds in noise and thresholds in quiet, which we refer to as thresholds-in-noise residual, was used as the outcome measure for the model. Explanatory measures were as follows: (1) auditory brainstem response (ABR) waves I and V amplitudes; (2) electrocochleographic action potential and summating potential amplitudes; (3) distortion product otoacoustic emissions level; and (4) categorical loudness scaling. All measurements were made at two frequencies (1 and 4 kHz). ABR and electrocochleographic measurements were made at 80 and 100 dB peak equivalent sound pressure level, while wider ranges of levels were tested during distortion product otoacoustic emission and categorical loudness scaling measurements. A model relating the thresholds-in-noise residual and the explanatory measures was created using multiple linear regression analysis.

RESULTS

Predictions of thresholds-in-noise residual using the model accounted for 61% (p < 0.01) and 48% (p < 0.01) of the variance in the measured thresholds-in-noise residual at 1 and 4 kHz, respectively.

CONCLUSIONS

Measures of thresholds in noise, the summating potential to action potential ratio, and ABR waves I and V amplitudes may be useful for the prediction of HHL in humans. With further development, our approach of quantifying HHL by the variance that remains in suprathreshold measures of auditory function after removing the variance due to thresholds in quiet, together with our statistical modeling, may provide a quantifiable and verifiable estimate of HHL in humans with normal hearing and with hearing loss. The current results are consistent with the view that inner hair cell and auditory nerve pathology may underlie suprathreshold auditory performance.

Influence of Instantaneous Compression on Recognition of Speech in Noise with Temporal Dips

BACKGROUND

In listening environments with background noise that fluctuates in level, listeners with normal hearing can "glimpse" speech during dips in the noise, resulting in better speech recognition in fluctuating noise than in steady noise at the same overall level (referred to as masking release). Listeners with sensorineural hearing loss show less masking release. Amplification can improve masking release but not to the same extent that it does for listeners with normal hearing.

PURPOSE

The purpose of this study was to compare masking release for listeners with sensorineural hearing loss obtained with an experimental hearing-aid signal-processing algorithm with instantaneous compression (referred to as a suppression hearing aid, SHA) to masking release obtained with fast compression. The suppression hearing aid mimics effects of normal cochlear suppression, i.e., the reduction in the response to one sound by the simultaneous presentation of another sound.

RESEARCH DESIGN

A within-participant design with repeated measures across test conditions was used.

STUDY SAMPLE

Participants included 29 adults with mild-to-moderate sensorineural hearing loss and 21 adults with normal hearing.

INTERVENTION

Participants with sensorineural hearing loss were fitted with simulators for SHA and a generic hearing aid (GHA) with fast (but not instantaneous) compression (5 ms attack and 50 ms release times) and no suppression. Gain was prescribed using either an experimental method based on categorical loudness scaling (CLS) or the Desired Sensation Level (DSL) algorithm version 5a, resulting in a total of four processing conditions: CLS-GHA, CLS-SHA, DSL-GHA, and DSL-SHA.

DATA COLLECTION

All participants listened to consonant-vowel-consonant nonwords in the presence of temporally-modulated and steady noise. An adaptive-tracking procedure was used to determine the signal-to-noise ratio required to obtain 29% and 71% correct. Measurements were made with amplification for participants with sensorineural hearing loss and without amplification for participants with normal hearing.

ANALYSIS

Repeated-measures analysis of variance was used to determine the influence of within-participant factors of noise type and, for participants with sensorineural hearing loss, processing condition on masking release. Pearson correlational analysis was used to assess the effect of age on masking release for participants with sensorineural hearing loss.

RESULTS

Statistically significant masking release was observed for listeners with sensorineural hearing loss for 29% correct, but not for 71% correct. However, the amount of masking release was less than masking release for participants with normal hearing. There were no significant differences among the amplification conditions for participants with sensorineural hearing loss.

CONCLUSIONS

The results suggest that amplification with either instantaneous or fast compression resulted in similar masking release for listeners with sensorineural hearing loss. However, the masking release was less for participants with hearing loss than it was for those with normal hearing.

Loudness Perception of Pure Tones in Parkinson's Disease

PURPOSE

Reduced intensity is a hallmark of speech production in Parkinson's disease (PD). Previous work has examined the perception of intensity in PD to explain these speech deficits. This study reports loudness ratings of pure tones by individuals with PD and controls, all with normal thesholds for older adults.

METHOD

Twenty individuals with PD and 23 age- and sex-matched controls rated the loudness of pure tones from 1 (very soft) to 7 (uncomfortably loud). Tones at 500, 750, 1000, 2000, and 4000 Hz were presented from 35 to 80 dB HL (or until a rating of 7 was given). A mixed-model analysis of variance was performed on ratings to assess the effects of group, frequency, sound intensity, and ear. Loudness growth slopes were determined for each participant and analyzed by group.

RESULTS

The mean loudness growth slopes of the control and PD groups did not significantly differ.

CONCLUSIONS

No difference was found in loudness growth slopes in response to externally generated tones in PD. This is in contrast with the findings of previous studies of self-generated speech and externally presented speech. The underlying causes for impaired perception and production of loudness in PD require further investigation.

Influence of suppression on restoration of spectral loudness summation in listeners with hearing loss

Loudness depends on both the intensity and spectrum of a sound. Listeners with normal hearing perceive a broadband sound as being louder than an equal-level narrowband sound because loudness grows nonlinearly with level and is then summed across frequency bands. This difference in loudness as a function of bandwidth is reduced in listeners with sensorineural hearing loss (SNHL). Suppression, the reduction in the cochlear response to one sound by the simultaneous presentation of another sound, is also reduced in listeners with SNHL. Hearing-aid gain that is based on loudness measurements with pure tones may fail to restore normal loudness growth for broadband sounds. This study investigated whether hearing-aid amplification that mimics suppression can improve loudness summation for listeners with SNHL. Estimates of loudness summation were obtained using measurements of categorical loudness scaling (CLS). Stimuli were bandpass-filtered noises centered at 2 kHz with bandwidths in the range of 0.1-6.4 kHz. Gain was selected to restore normal loudness based on CLS measurements with pure tones. Gain that accounts for both compression and suppression resulted in better restoration of loudness summation, compared to compression alone. However, restoration was imperfect, suggesting that additional refinements to the signal processing and gain-prescription algorithms are needed.

Deriving loudness growth functions from categorical loudness scaling data

The goal of this study was to reconcile the differences between measures of loudness obtained with continuous, unbounded scaling procedures, such as magnitude estimation and production, and those obtained using a limited number of discrete categories, such as categorical loudness scaling (CLS). The former procedures yield data with ratio properties, but some listeners find it difficult to generate numbers proportional to loudness and the numbers cannot be compared across listeners to explore individual differences. CLS, where listeners rate loudness on a verbal scale, is an easier task, but the numerical values or categorical units (CUs) assigned to the points on the scale are not proportional to loudness. Sufficient CLS data are now available to assign values in sones, a scale proportional to loudness, to the loudness categories. As a demonstration of this approach, data from Heeren, Hohmann, Appell, and Verhey [J. Acoust. Soc. Am. 133, EL314-EL319 (2013)] were used to develop a CUsone metric, whose values were then substituted for the original CU values in reanalysis of a large set of CLS data obtained by Rasetshwane, Trevino, Gombert, Liebig-Trehearn, Kopun, Jesteadt, Neely, and Gorga [J. Acoust. Soc. Am. 137, 1899-1913 (2015)]. The resulting data are well fitted by power functions and are in general agreement with previously published results obtained with magnitude estimation, magnitude production, and cross modality matching.

Relative contributions of specific frequency bands to the loudness of broadband sounds

Listeners with normal hearing (NH) and sensorineural hearing loss (SNHL) were asked to compare pairs of noise stimuli and choose the louder noise in each pair. Each noise was made up of 15, two-ERB_N (equivalent rectangular bandwidth) wide frequency bands that varied independently over a 12-dB range from one presentation to the next. Mean levels of the bands followed the long-term average speech spectrum (LTASS) or were set to 43, 51, or 59 dB sound pressure level (SPL). The relative contribution of each band to the total loudness of the noise was determined by computing the correlation between the difference in levels for a given band on every trial and the listener's decision on that trial. Weights for SNHL listeners were governed by audibility and the spectrum of the noise stimuli, with bands near the spectral peak of the LTASS noise receiving greatest weight. NH listeners assigned greater weight to the lowest and highest bands, an effect that increased with overall level, but did not assign greater weight to bands near the LTASS peak. Additional loudness-matching and paired-comparison studies using stimuli missing one of the 15 bands showed a significant contribution by the highest band, but properties other than loudness may have contributed to the decisions.

Development of a multi-category psychometric function to model categorical loudness measurements

A multi-category psychometric function (MCPF) is introduced for modeling the stimulus-level dependence of perceptual categorical probability distributions. The MCPF is described in the context of individual-listener categorical loudness scaling (CLS) data. During a CLS task, listeners select the loudness category that best corresponds to their perception of the presented stimulus. In this study, CLS MCPF results are reported for 37 listeners (15 normal hearing, 22 with hearing loss). Individual-listener MCPFs were parameterized, and a principal component analysis (PCA) was used to identify sources of inter-subject variability and reduce the dimensionality of the data. A representative "catalog" of potential listener MCPFs was created from the PCA results. A method is introduced for using the MCPF catalog and maximum-likelihood estimation, together, to derive CLS functions for additional participants; this technique improved the accuracy of the CLS results and provided a MCPF model for each listener. Such a technique is particularly beneficial when a relatively low number of measurements are available (e.g., International Standards Organization adaptive-level CLS testing). In general, the MCPF is a flexible tool that can characterize any type of ordinal, level-dependent categorical data. For CLS, the MCPF quantifies the suprathreshold variability across listeners and provides a model for probability-based analyses and methods.