Development of a virtual bimodal Cochlea Implant listener
Final Report Abstract
The goal of the present DFG-project was to develop models and simulation tools to predict spatial speech-in-noise performance of cochlear implant (CI) users with contralateral acoustic hearing, i.e., bimodal CI listeners. Therefore, a four-step procedure was taken: In the first step, speech-in-noise performance of 16 actual CI patients was measured using a novel mobile measurement setup incorporating the use of virtual acoustics to create spatial acoustic scenes. This setup also included the Master Hearing Aid to replace the patient’s own hearing aid and cable transmission to the patient’s own CI speech processor. The setup allowed for control of head movements, control for different hearing aid settings, and control for highly reproducible spatial scenes. Speech reception thresholds (SRTs) were measured using this setup with bimodal CI listeners, as well as with CI-SSD users (patients, who were implanted after single-sided deafness, i.e., who consist of normal hearing contralateral to their CI). The resulting dataset was used as a baseline for comparison of subsequent simulation results and model results in the next steps. In the second step, simulation tools (“vocoders”) of CI listening and listening with impaired acoustic hearing have been developed and tested using normal-hearing listeners with exactly the same controlled conditions as in the first step. The speech performance results of the simulations showed a good reproduction of bimodal CI user’s speech performance. In the third step, the simulation tools of the previous second step have been used to develop a computer model that predicts SRTs for bimodal CI and CI-SSD listeners. Here, an established “normal” binaural processing was used, which is in principle capable to effectively process right and left signal for the purpose of enhancing target speech. With this first version of the virtual bimodal CI listener model, a good reproduction of CI-SSD and bimodal CI users’ SRTs was observed. Interesting here was that the model, although in principle capable of doing binaural processing, was not using any binaural processing to enhance target speech and attenuate other signals. The reason for that was investigated and was found in missing fine-grained timing information due to the CI signal processing. This was confirmed in an additional experiment with bilateral CI users. In the fourth step, two versions of a “microscopic” model of bimodal CI user’s spatial speech-in-noise performance were developed, which use automatic speech recognition techniques on speech features that were subject to electric and acoustic processing, as in CI-SSD and bimodal CI users. These model versions used either a simple “better-earlistening” hypothesis or an advanced interaction in the form of time-alignment of acoustic and electric features. The results showed that the better-ear-listening model version was providing a satisfactory reproduction of SRTs, which underlined the observation from the measurements in the first step. The simulation tools have been used to evaluate state-of-the-art preprocessing algorithms for signal enhancement that were also evaluated with actual bimodal CI patients. Both, the CI simulation, as well as the model version of the third step was made publicly available for further research studies, as well as for product-development.
Publications
-
(2017) “Spatial release from masking in actual and simulated bimodal and single-sided deaf cochlear implant listeners,” Conference on implantable auditory prostheses (CIAP), Lake Tahoe, CA, USA
Williges, B., Wesarg, T., Jung, L., Geven, L., Radeloff, A., Jürgens, T.
-
(2018) „Räumliches Sprachverstehen in bimodalen und einseitig ertaubten Cochlea-Implantatträgern: Messungen und physiologische Modellierung“, 21. Jahrestagung der Deutschen Gesellschaft für Audiologie (DGA), Halle an der Saale
Williges, B., Wesarg, T., Jung, L., Geven, L., Radeloff, A., Jürgens, T.
-
(2018). "Coherent Coding of Enhanced Interaural Cues Improves Sound Localization in Noise With Bilateral Cochlear Implants." Trends in Hearing 22: 1-15
Williges, B., Jürgens, T., Hu, H., Dietz, M.
-
(2018). "Modeling Speech Intelligibility of Simulated Bimodal and Single-Sided Deaf Cochlear Implant Users." Acta acustica united with Acustica 104: 918-921
Zedan, A., Williges, B., Jürgens, T.
-
(2019). “Spatial speech-in-noise performance in bimodal and single-sided deaf cochlear implant users I: Assessment,“ Trends in Hearing
Williges, B., Wesarg, T., Jung, L., Geven, L., Radeloff, A., Jürgens, T.
