12th Speech in Noise Workshop, 9-10 January 2020, Toulouse, FR

Listening effort in normal-hearing listeners with a cochlear implant vocoder simulation using subjective ratings and EEG measurements

Angelika Kothe
Jade University of Applied Sciences, Oldenburg, Germany | Fraunhofer IDMT-HSA, Oldenburg, Germany

Amy J. Hall(b)
Fraunhofer IDMT-HSA, Oldenburg, Germany

Michael Schulte
Hörzentrum Oldenburg GmbH, Oldenburg, Germany

Kamil Adiloglu
HörTech gGmbH, Oldenburg, Germany

Axel H. Winneke(a)
Fraunhofer IDMT, Oldenburg, Germany

(a) Presenting
(b) Attending

Many hearing impaired individuals experience understanding speech in noisy environments (SpiN) to be exhausting. Being exposed to a challenging listening situation for a prolonged period of time can cause fatigue. This can lead to a chronic burden, which is a disadvantage in everyday life and in the workplace.

In the context of hearing devices and listening effort, less research has been conducted with cochlear implant (CI) patients. This motivated the current study to look more closely at how listening to SpiN with a CI affects listening effort. Since recruitment of sufficiently large number of CI users is more difficult to realize and because of the heterogeneity of the CI patients, in this project we equipped normal hearing participants with behind-the-ear (BTE) hearing aid dummies and a vocoder (Bräker et al., 2009, Z. Audiol. 48:158) to acoustically simulate a CI. Participants perform the adaptive categorical listening effort scaling task (ACALES, Krüger et al., 2017, JASA 141:4680) to obtain subjective ratings on experienced listening effort at various SNRs. A multi-channel electroencephalogram (EEG) is recorded simultaneously in order to quantify listening effort on a neurophysiological level. The EEG parameter of interest here are changes in the alpha-frequency band. The study investigates the effect of spectral and temporal degradation of a speech signal (i.e. CI simulation) in noise as well as the effect of noise suppression on listening effort. Noise suppression is performed using a minimum variance distortionless response (MVDR) beam former to reduce the impact of a noise source located behind the participant at 135° on the speech source at 0°. The whole signal processing chain including the noise suppression and vocoder is implemented in the real-time capable master hearing-aid (MHA, Grimm et al., 2006, Acta. Acust. united Ac. 92:618) platform and all measurements are conducted using real-time audio.

Despite the limitations of a simulation study, the project will shed light on listening effort and CIs, how listening effort is reflected in the EEG and how spatial noise suppression can aid CI users in terms of reducing listening effort when listening to SpiN.

Last modified 2020-01-06 19:23:55