Johan J. Hanekom

Sentence recognition in noise: Variables in compilation and interpretation of tests.

Abstract Tests of sentence recognition in noise constitute an essential tool for the assessment of auditory abilities that are representative of everyday listening experiences. A number of recent articles have reported on the development of such tests, documenting different approaches and methods. However, both the development and interpretation of these tests require careful consideration of many variables. This article reviews and categorizes the stimulus, presentation, subject, response, and performance variables influencing the development and interpretation of tests of sentence recognition in noise. A systematic framework is utilized to document published findings on these variables. Recommendations and guidelines, based on test performance requirements and test objectives, are provided concerning the interpretation of results and the development of new test materials. Sumario La ECoG ha mostrado desde hace mucho que complementa el diagnóstico de la MD, primariamente por la medición de la tasa de amplitud SP/AP. Aunque se reporta en la literatura como una prueba de alta especificidad para este trastorno, la sensibilidad de la ECoG, en la población general de MD, se mantiene relativamente baja (rango de 20-65%), El presente estudio evaluó la sensibilidad y la especificidad del protocolo de ECoG que empleamos para pacientes con sospecha de MD, que incluyó la medición de la amplitud y las áreas de los SP y AP ante clicks (para derivar las tasas de amplitud y de área de SP/AP) y la amplitud del SP ante bursts tonales de 1000 y 2000 Hz. Se condujo un cuadro de revisión retrospectivo para comparar los resultados de de la ECoG de 178 pacientes sospechosos de MD, con su diagnóstico eventual. Las medidas de la mayor sensibilidad y especificidad (determinadas utilizando un análisis logístico de regresión) incluyeron: amplitud del PS, área del PS, tasa de área SP/AP y área total de SP-AP. Los valores de sensibilidad y especificidad asociados con esas medidas fueron de 92% y 84%, respectivamente. El valor de sensibilidad fue considerablemente mayor que el previamente reportado y esto es atribuible a la inclusión de mediciones de área en nuestro protocolo.

Free field frequency discrimination abilities of cochlear implant users.

Poor music perception abilities of cochlear implant users may be attributed to limited pitch resolution afforded by the implant system. We investigated (i) what the typical frequency discrimination thresholds of cochlear implant users would be in free field listening conditions and (ii) whether frequency discrimination behaviour would be influenced by the position of the reference frequency relative to the frequency response of filters selected from the users map. Frequency discrimination thresholds were determined according to an adaptive two-alternative forced choice (2AFC) method, using pure tones delivered in free field conditions. Results showed that finer frequency resolution than previously thought could be available to cochlear implant users. Results are interpreted in terms of intermediate pitch percepts possibly created by near-simultaneous activation of adjacent electrodes, resulting in overlapping neural populations to be stimulated. The findings may contribute to strategies aiming to improve music perception abilities of cochlear implant users.

Can subject-specific single-fibre electrically evoked auditory brainstem response data be predicted from a model?

This article investigates whether prediction of subject-specific physiological data is viable through an individualised computational model of a cochlear implant. Subject-specific predictions could be particularly useful to assess and quantify the peripheral factors that cause inter-subject variations in perception. The results of such model predictions could potentially be translated to clinical application through optimisation of mapping parameters for individual users, since parameters that affect perception would be reflected in the model structure and parameters. A method to create a subject-specific computational model of a guinea pig with a cochlear implant is presented. The objectives of the study are to develop a method to construct subject-specific models considering translation of the method to in vivo human models and to assess the effectiveness of subject-specific models to predict peripheral neural excitation on subject level. Neural excitation patterns predicted by the model are compared with single-fibre electrically evoked auditory brainstem responses obtained from the inferior colliculus in the same animal. Results indicate that the model can predict threshold frequency location, spatial spread of bipolar and tripolar stimulation and electrode thresholds relative to one another where electrodes are located in different cochlear structures. Absolute thresholds and spatial spread using monopolar stimulation are not predicted accurately. Improvements to the model should address this.

The relative importance of spectral cues for vowel recognition in severe noise

The importance of formants and spectral shape was investigated for vowel perception in severe noise. Twelve vowels were synthesized using two different synthesis methods, one where the original spectral detail was preserved, and one where the vowel was represented by the spectral peaks of the first three formants. In addition, formants F1 and F2 were suppressed individually to investigate the importance of each in severe noise. Vowels were presented to listeners in quiet and in speech-shaped noise at signal to noise ratios (SNRs) of 0, -5, and -10 dB, and vowel confusions were determined in a number of conditions. Results suggest that the auditory system relies on formant information for vowel perception irrespective of the SNR, but that, as noise increases, it relies increasingly on more complete spectral information to perform formant extraction. A second finding was that, while F2 is more important in quiet or low noise conditions, F1 and F2 are of similar importance in severe noise.

Threshold predictions of different pulse shapes using a human auditory nerve fibre model containing persistent sodium and slow potassium currents.

The ability of a human auditory nerve fibre computational model to predict threshold differences for biphasic, pseudomonophasic and alternating monophasic waveforms was investigated. The effect of increasing the interphase gap, interpulse interval and pulse rate on thresholds was also simulated. Simulations were performed for both anodic-first and cathodic-first stimuli. Results indicated that the model correctly predicted threshold reductions for pseudomonophasic compared to biphasic waveforms, although reduction for alternating monophasic waveforms was underestimated. Threshold reductions were more pronounced for cathodic-first stimuli compared to anodic-first stimuli. Reversal of the phases in pseudomonophasic stimuli suggested a threshold reduction for anodic-first stimuli, but a threshold increase in cathodic-first stimuli. Inclusion of the persistent sodium and slow potassium currents in the model resulted in a reasonably accurate prediction of the non-monotonic threshold behaviour for pulse rates higher than 1000 pps. However, the model did not correctly predict the threshold changes observed for low pulse rate biphasic and alternating monophasic waveforms. It was suggested that these results could in part be explained by the difference in the refractory periods between real and simulated auditory nerve fibres, but also by the lack of representation of stochasticity observed in real auditory nerve fibres in our auditory nerve model.

Estimation of stimulus attenuation in cochlear implants

Neural excitation profile widths at the neural level, for monopolar stimulation with Nucleus straight and contour arrays respectively, were simulated using a combined volume-conduction-neural model. The electrically evoked compound action potential profile widths at the electrode array level were calculated with a simple approximation method employing stimulus attenuation inside the cochlear duct, and the results compared to profile width data from literature. The objective of the article is to develop a simple method to estimate stimulus attenuation values by calculating the values that best fit the modelled excitation profile widths to the measured evoked compound action potential profile widths. Results indicate that the modelled excitation profile widths decrease with increasing stimulus attenuation. However, fitting of modelled excitation profile widths to measured evoked compound action potential profile widths show that different stimulus attenuation values are needed for different stimulation levels. It is suggested that the proposed simple model can provide an estimate of stimulus attenuation by calculating the value of the parameter that produces the best fit to experimental data in specific human subjects.

Constructing a three-dimensional electrical model of a living cochlear implant user's cochlea.

BACKGROUND Hearing performance varies greatly among users of cochlear implants. Current three-dimensional cochlear models that predict the electrical fields inside a stimulated cochlea and their effect on neural excitation are generally based on a generic human or guinea pig cochlear shape that does not take inter-user morphological variations into account. This precludes prediction of user-specific performance. AIMS The aim of this study is to develop a model of the implanted cochlea of a specific living human individual and to assess if the inclusion of morphological variations in cochlear models affects predicted outcomes significantly. METHODS Five three-dimensional electric volume conduction models of the implanted cochleae of individual living users were constructed from standard CT scan data. These models were embedded in head models that include monopolar return electrodes in accurate anatomic positions. Potential distributions and neural excitation patterns were predicted for each of the models. RESULTS Modeled potential distributions and neural excitation profiles (threshold amplitudes, center frequencies, and bandwidths) are affected by user-specific cochlear morphology and electrode placement within the cochlea. CONCLUSIONS This work suggests that the use of user-specific models is indicated when more detailed analysis is required than what is available from generic models. Copyright

The performance of different synthesis signals in acoustic models of cochlear implants

Synthesis (carrier) signals in acoustic models embody assumptions about perception of auditory electric stimulation. This study compared speech intelligibility of consonants and vowels processed through a set of nine acoustic models that used Spectral Peak (SPEAK) and Advanced Combination Encoder (ACE)-like speech processing, using synthesis signals which were representative of signals used previously in acoustic models as well as two new ones. Performance of the synthesis signals was determined in terms of correspondence with cochlear implant (CI) listener results for 12 attributes of phoneme perception (consonant and vowel recognition; F1, F2, and duration information transmission for vowels; voicing, manner, place of articulation, affrication, burst, nasality, and amplitude envelope information transmission for consonants) using four measures of performance. Modulated synthesis signals produced the best correspondence with CI consonant intelligibility, while sinusoids, narrow noise bands, and varying noise bands produced the best correspondence with CI vowel intelligibility. The signals that performed best overall (in terms of correspondence with both vowel and consonant attributes) were modulated and unmodulated noise bands of varying bandwidth that corresponded to a linearly varying excitation width of 0.4 mm at the apical to 8 mm at the basal channels.

Perception of vowels and prosody by cochlear implant recipients in noise.

UNLABELLED The aim of the present study was to compare the ability of cochlear implant (CI) recipients to recognise speech prosody in the presence of speech-weighted noise to their ability to recognise vowels in the same test paradigm and listening condition. All test materials were recorded from four different speakers (two male, two female). Two prosody recognition tasks were developed, both using single words as stimuli. The first task involved a question/statement distinction, while the second task required listeners to make a judgement about the speakers attitude. Vowel recognition tests were conducted using vowel pairs selected on the basis of specific acoustic cues (frequencies of the first two formants and duration). Ten CI users and ten normal-hearing controls were tested in both quiet and an adaptive noise condition, using a two-alternative forced-choice test paradigm for all the tests. Results indicated that vowel recognition was significantly better than prosody recognition in both listener groups in both quiet and noise, and that question/statement discrimination was the most difficult task for CI listeners in noise. Data from acoustic analyses were used to interpret differences in performance on different tasks and with different speakers. LEARNING OUTCOMES As a result of this activity, readers will be able to (1) describe suitable methods for comparing vowel and prosody perception in noise, (2) compare performance on vowel and prosody perception tasks in quiet in normal-hearing listeners and cochlear implant recipients, (3) compare performance on vowel and prosody perception tasks in noise in normal-hearing listeners and cochlear implant recipients and (4) relate performance on prosody tasks in quiet to performance on these tasks in noise.

An analysis of the effects of electrical field interaction with an acoustic model of cochlear implants

Electrical field interaction caused by current spread in a cochlear implant was modeled in an explicit way in an acoustic model (the SPREAD model) presented to six listeners with normal hearing. The typical processing of cochlear implants was modeled more closely than in traditional acoustic models by careful selection of parameters related to current spread or parameters that could amplify the electrical field interactions caused by current spread. These parameters were the insertion depth, electrode spacing, electrical dynamic range, and dynamic range compression function. The hypothesis was that current spread could account for the asymptote in performance in speech intelligibility experiments observed at around seven stimulation channels in a number of cochlear implant studies. Speech intelligibility for sentences, vowels, and consonants at three noise levels (SNR of +15 dB, +10 dB, and +5 dB) was measured as a function of the number of spectral channels (4, 7, and 16). The SPREAD model appears to explain the asymptote in speech intelligibility at seven channels for all noise levels for all speech material used in this study. It is shown that the compressive amplitude mapping used in cochlear implants can have a detrimental effect on the number of effective channels.