Sennheiser E 945 - 03-06 User Manual download pdf (Page 82)

Deutsche Gesellschaft für auDioloGie e. V.162 163

AbstrActs

18. JAhrestAgung „MultidisziplinArität in der Audiologie“

Beitrag wird präsentiert am 05.03.2015 um 16.54 Uhr im Rahmen der SS03

What are the requirements for modern hearing aid algorithms? Consequences from

state-of-the-art speech intelligibility models for hearing-impaired listeners

T. Brand

Cluster of Excellence „Hearing4All“ and Medical Physics, University of Oldenburg

The optimization of speech intelligibility is an important goal for developing, optimizing, and

ﬁtting of hearing aid algorithms. Speech intelligibility tests with human listeners are indispens-

able for evaluating the ﬁnal result. Nevertheless, during the optimization process intelligibility

models can be used to make predictions about the ﬁnal outcome.

Traditional models like the Speech-Intelligibility-Index (SII) analyze the SNR in different fre-

quency bands. According to these auditory models, hearing aids should restore audibility

of speech as far as possible. Some authors use these models also for estimating the role

of dead regions and to predict whether or not ampliﬁcation should be applied in these fre-

quency regions. Short-term SII versions can furthermore predict release from masking due to

modulations of the interfering noise. Other models like the Speech-Transmission-Index (STI),

the Envelope-Power-Spectrum-Model (EPSM), and Short-Time Objective Intelligibility (STOI)

analyze modulations and can make predictions about dynamic compression, noise reduction

algorithms, and also signal distortions. Different studies compared the performance of such

models for different acoustic scenarios.

The release from masking due to binaural processing in listeners with impaired hearing can

be predicted very well by models based on the Equalization-Cancellation (EC) theory. These

models can make predictions about the restauration of the listener’s binaural processing using

hearing aids and about the consequence of binaural coupling of dynamic compression.

Auditory models are able to describe speciﬁc aspects of hearing impairment. Such models

can be combined with automatic speech recognizers enabling predictions about intelligibility

as well.

Beitrag wird präsentiert am 05.03.2015 um 18.06 Uhr im Rahmen der SS03

Bimodal auditory prostheses: algorithm development and evaluation

T. Francart, J. Wouters

ExpORL, Dept. Neurosciences, KU Leuven, Belgium

When a cochlear implant (CI) recipient has residual hearing in the non-implanted ear, it can be

acoustically stimulated using a hearing aid together with the CI. This has been shown to be

advantageous compared to only CI stimulation. However, while the two ears are stimulated,

with current clinical devices bimodal listeners do not appear to use binaural cues, due to poor

cue transmission by the devices and inappropriate ﬁtting. We present two sound processing

strategies to improve this: MEnS and SCORE bimodal.

The SCORE bimodal strategy aims to equalise loudness growth for both modalities using real-

time application loudness models. SCORE was evaluated using loudness balancing, speech

perception and sound source localisation experiments with 6 bimodal listeners. As SCORE

was found not to interfere with speech perception and to improve loudness perception, it

seems beneﬁcial to implement it in clinical devices. To better individualise the loudness mod-

els, we are developing a method based on auditory steady state responses in the EEG to

measure loudness growth functions with electrical and acoustical stimulation.

The MEnS strategy was developed to improve perception of interaural time differences (ITDs)

with bimodal stimulation. It introduces temporal modulations on all stimulated electrodes,

synchronously with modulations present in the acoustic signal. It was evaluated by measuring

ITD sensitivity and extent of lateralisation. Compared to the ACE strategy, performance was

signiﬁcantly improved with just noticeable differences well within the physically relevant range.

The application of these two strategies could lead to improved sound source localisation and

binaural unmasking.

Literatur:

Maaike Van Eeckhoutte was supported by a PhD-grant for Strategic Basic Research by the Agency for Innovation by

Science and Technology in Flanders (IWT, 131106).

Dimitar Spirrov was supported by an IWT Baekeland grant from the Agency for Innovation by Science and Technology

in Flanders together with Cochlear (IWT 140748).

1 2 ... 77 78 79 80 81 82 83 84 85 86 87 ... 155 156

Comments to this Manuals

No comments

Sennheiser E 945 - 03-06 User Manual Page 82

Comments to this Manuals

Related products and manuals for Unknown Sennheiser E 945 - 03-06