9:00 |
-
Spatial Perception and Speech Intelligibility in Aided Listeners
Jens Cubick, Jörg Buchholz, Virginia Best, Mathieu Lavandier, Torsten Dau
[Abstract]
Cubick and Dau (2016) showed that speech reception
thresholds (SRTs) in noise, obtained with normal-hearing
(NH) listeners, can
be significantly higher with hearing aids (HAs) than in the
corresponding unaided condition. Some of the listeners
reported a
change in their spatial perception of the sounds due to the
HA processing, with auditory images often being broader
and closer to
the listener or even internalized. The current study
investigated whether worse speech intelligibility with HAs
might be caused by a
”shrunken” acoustic scene and thus a reduced ability to
spatially separate the target speech from the interferers.
SRTs were
measured in NH listeners with and without ”ideal” HAs
(broadband, linear, flat gain) in the presence of three
interfering talkers
or speech-shaped noises. The interferers were presented
either at +/- 90 and 180 degrees azimuth or were co-located
with the
target sentence at 0 degrees. Consistent with the previous
study, SRTs were found to be increased by 2-2.5 dB with
HAs when the
interferers were spatially separated, but only by 0.5-1 dB
when they were co-located. The 1.5 dB difference indicates
that at least
some of the disruption to speech intelligibility caused by HAs
can potentially be attributed to degraded spatial separation.
|
9:20 |
-
On the relation between hearing impairment, psychoacoustic measurement outcomes and speech recognition tasks: How far do we get with the FADE model?
David Hülsmeier, Marc René Schädler, Mareike Buhl, Anna Warzybok, Birger Kollmeier
[Abstract]
Psychoacoustic modeling and human and machine speech recognition research are still separate research fields, with different approaches and goals. As a unifying approach, the simulation framework for auditory discrimination experiments (FADE) was proposed, which was shown to successfully model the behavior of normal-hearing listeners in basic psychoacoustic experiments and speech recognition tasks based on automatic speech recognition (ASR) techniques. To better generalize, only basic assumptions about the signal representation (auditory signal processing) and the detection process (ASR back-end) are made. This approach is extended here towards hearing impairment by the inclusion of different effective auditory deficiencies such as the absolute hearing threshold, a supra-threshold distortion, and a reduced spectral resolution. The respective effect of these effective signal processing deficiencies on the outcome of typical measures is evaluated, i.e. on tone detection in quiet or in (notched-) noise, and on speech recognition tasks with different noise types. The simulated outcomes suggest that it is not possible to reliably characterize different auditory deficiencies using only one type of psychoacoustic task. FADE could possibly be used to infer a functional description of the impaired auditory system from a range of (psycho-)acoustic experiments and to individually predict outcomes of speech recognition tests.
|
9:40 |
-
Modeling Refractoriness In Phenomenological Models of Electrically-Stimulated Auditory Nerve Fibers
Marko Takanen, Jean-Noël Weller, Bernhard Seeber
[Abstract]
Functionality of cochlear implants is based on electrical signals charging the capacitive cell membrane of the auditory nerve fiber to push the membrane voltage above the neuron’s firing threshold. The neuron then spikes shortly after the threshold crossing and, therefore, there is always a stochastic delay, called latency, between the onset of the electrical pulse and the time of spiking. After spiking, the neuron exhibits refractoriness and cannot be excited during the absolute refractory period, after which it gradually recovers to the resting state. In refractoriness, it is important to consider also the latency of the neuron to the preceding excitatory pulse. Unfortunately, neurophysiological measurements for refractoriness have ignored the latency to the supra-threshold conditioning pulse; resulting in excessively large estimates for the absolute refractory period. Consequently, several existing models, modeling both latency and refractoriness using time constants from such measurements, end up overestimating the absolute refractory period. Here, we present a solution that considers the estimated latency of the neuron when adjusting the duration of the absolute refractory period after each spike, allowing our model to reproduce both latency and refractoriness data from literature, and improving the model’s accuracy in predicting responses to high-pulse-rate stimulation.
Supported by BMBF 01GQ1004B.
|
10:00 |
-
Untersuchungen zum Einfluss raumakustischer Parameter auf die Hörwahrnehmung bei Cochlea-Implantat-Trägern
Bernhard Eurich, Michael Oehler, Thomas Klenzner
[Abstract]
Die Hörwahrnehmung von Cochlea-Implantat-Trägern wird durch diverse Einflussfaktoren erheblich erschwert (Capretta, Nittrouer & Moberly, 2015).
Eine wesentliche Rolle spielen in diesem Zusammenhang häufig die raumakustischen Begebenheiten. Selbst bei Personen ohne Hörbeeinträchtigung
sind raumakustische Faktoren für die Wahrnehmung der auditiven Szene (z.B. hinsichtlich Sprachverständlichkeit oder Hörsamkeit) von großer
Bedeutung (Cook et al., 2013; Meyer 1995; Zahorik & Brandewie, 2016). Bei CI-Trägern sind komplexe Hörumgebungen meist mit besonders starken
Einbußen verbunden (Igelhart, 2016; Loiselle et al., 2016; Zeitler et al., 2015; Roy et al., 2015).
Ziel der Studie ist es, jene raumakustischen Kriterien und deren Ausprägung zu ermitteln, die auf die Hörsamkeit und Verständlichkeit bei CI-Trägern
besonderen Einfluss nehmen. Hierzu wurden sowohl Sprachsignale aus dem Oldenburger Satztest (OLSA) als auch einfache musikalische Stimuli mit
der Software Auratorium auralisiert. Dabei dienten normgemäße kleine bis mittelgroße Räume (DIN 18041:2016-03) für sprachliche Kommunikation bzw.
Musikdarbietung als Grundlage. Die räumlichen Eigenschaften wurden hinsichtlich Primär- und Sekundärstrukturen bzw. Absorptionsverhalten der
Oberflächen variiert.
Die erstellten Stimuli werden 15 Probanden über den Audio-Eingang ihres Cochlea-Implantats präsentiert. Mittels eines adaptierten MUSHRA-Tests
sollen Verständlichkeitsschwellen und Hörsamkeitsbewertungen ermittelt werden. Die gewonnenen Aussagen sollen helfen, die Einflüsse bestimmter
raumakustischer Verhältnisse auf die Qualität des Hörerlebnisses bei CI-Trägern zu beurteilen und zu verbessern.
|
10:40 |
-
Messtechnischer Nachweis der mehrkanaligen Signalverarbeitung bei Hörgeräten
Hendrik Husstedt, Anne Wiggers
[Abstract]
Die Signalverarbeitung in heutigen Hörsystemen geschieht in der Regel in verschiedenen Kanälen. Dabei wird im Bereich der Hörsysteme unter einem Kanal die Möglichkeit verstanden, dass mindestens die Verstärkung und die Kompression der Lautstärke in einem gewissen Frequenzbereich separat eingestellt werden kann. Für die Anpassung von Hörsystemen ist diese Eigenschaft von besonderer Bedeutung, sodass die Mehrkanaligkeit ein wichtiges Qualitätsmerkmal darstellt. Dies zeigt sich auch darin, dass in einigen Ländern eine gewisse Mindestanzahl an Kanälen bei Hörsystemen gefordert wird, z.B. vier Kanäle in Deutschland. Obwohl die Mehrkanaligkeit eine wichtige Eigenschaft von Hörsystemen darstellt, existiert bislang weder eine standardisierte Definition noch steht ein einheitliches Verfahren für einen messtechnischen Nachweis zur Verfügung. Aus diesem Grund beschäftigt sich dieser Beitrag mit der Möglichkeit die Mehrkanaligkeit bei Hörsystemen messtechnisch nachzuweisen. Zu diesem Zweck werden zunächst Messergebnisse präsentiert, die die Auswirkungen der Mehrkanaligkeit anschaulich darstellen. Anhand dieser Ergebnisse wird diskutiert inwieweit sich voneinander abhängige und unabhängige Kanäle unterscheiden lassen. Schließlich wird gezeigt wie mit diesen Überlegungen die Anzahl der voneinander unabhängigen Kanäle messtechnisch nachgewiesen werden kann.
|
11:00 |
-
Development and Analysis of the Polish Speech Test Signal in View of the IEC 60118-15 Standard
Dorota Habasinska, Ewa Skrodzka
[Abstract]
The introduction of the International Speech Test Signal (ISTS) in 2012 made a breakthrough in the practice of hearing aids performance assessment. However, due to the limitation of its linguistic content, ISTS might not be representative of any language but the six used to create it. That is why its worldwide application in the domain of hearing aids fitting might raise some reservations. For instance, Polish native speakers might benefit from ISTS-based hearing aid fitting procedure less, compared to the same procedure employing a signal representative of Polish speech. This assumption served as a motivation for the hereby presented studies. This paper presents the development procedure of the Polish Speech test Signal (PSTS). PSTS is a counterpart of ISTS, though based only on Polish speech recordings and it is in principle representative of Polish speech. Multiple comparison analysis between PSTS and ISTS (including among others long-term average speech spectrum, percentile rank of short-term power levels, content of voiceless parts of speech) fall also within the scope of the paper.
|
11:20 |
-
The effect of dynamic range compression on spatial perception in a reverberant environment
Henrik Hassager, Alan Wiinberg, Torsten Dau
[Abstract]
The present study investigated the effects of fast-acting hearing-aid compression on normal-hearing and hearing-impaired listeners’ spatial perception in a reverberant environment. Three compression schemes - independent compression at each ear; linked compression between the two ears; and ”spatially ideal” compression operating solely on the dry source signal - were considered using virtualized speech and noise bursts. Listeners indicated the location and distribution of their perceived sound images on the horizontal plane graphically on a touch screen. A linear amplification scheme was considered as the reference condition. The results showed that both independent and linked compression resulted in more diffuse and broader sound images as well as internalization and image splits, whereby more image splits were reported for the noise bursts than for speech. Only the spatial ideal compression provided the listeners with a spatial percept similar to that obtained with linear processing. The same general pattern was observed for both listener groups. An analysis of the interaural cross-correlation and direct-to-reverberant ratio suggested that the spatial distortions resulted from enlarged reverberant energy. Thus, modifictions of the relation between the direct and the reverberant part of the sound should be avoided in amplification strategies that attempt to preserve the natural sound scene.
|
11:40 |
-
How should we model the supra-threshold distortion component of hearing impairment?
Marc René Schädler, David Hülsmeier, Anna Warzybok, Mareike Buhl, Birger Kollmeier
[Abstract]
Hearing-impaired listeners can show very different speech recognition performance despite identical audiograms, which is usually not found in normal-hearing listeners.
This remaining individual ”distortion” component, which cannot be explained by the loss of sensitivity was hypothesized to be due to different auditory processing deficiencies, cognitive factors, or simply ”age”.
This contribution explores the suitability of a level uncertainty to model the effect of individual supra-threshold signal processing deficiency.
It is assumed to be independent of the signal level and the individual sensitivity loss, and supposed to affect tone-in-noise detection thresholds.
To test the hypothesis, speech recognition performance in quiet, stationary, and fluctuating noise, and tone-in-noise detection thresholds were measured of 20 listeners with audiograms from normal-hearing to severe hearing loss.
The individual audiogram and/or the tone in noise detection thresholds are used to perform individualized predictions of the outcome of the matrix sentence test to assess their contribution in the different conditions.
Using level uncertainty as the cause for distortion loss can increase the prediction accuracy of speech recognition thresholds (recently shown using the established framework for auditory discrimination experiments; FADE).
Here, the questions of an appropriate implementation and a psychoacoustic measure to determine its degree are addressed.
|