Contribution
Testing hearing-aid features for challenging listening situations using the Concurrent Matrix Test
* Presenting author
Abstract:
The Concurrent Matrix Test is a multi-talker speech-test with overlapping sentences from three talkers and directions: During a continuous presentation of sentences, the test persons have to repeat words for alternating target talkers, which are to be identified through call-sign detection. The paradigm is motivated by the challenging listening situation of a group conversation in a restaurant with turn-taking and overlapping speech from different talkers. It allows for assessing speech recognition at high signal-to-noise ratios (SNR) from the three talker directions in a single measurement. This makes it particularly interesting for evaluating hearing-aid features, including noise-reduction algorithms with SNR dependencies and directional patterns in a highly relevant use case for hearing-impaired people. However, in such a complex listening situation with supra-threshold signal presentation, measuring an observable of interest is not trivial. Particularly, the individual cognitive speech-processing speed has turned out to be an important factor. When investigating benefits of hearing-aid features, the Concurrent Matrix Test can be used as a conclusive measure for complex listening situations, or to decompose the interplay of SNR, cognitive load, and spatial cues. In this contribution, the paradigm is introduced and results of hearing-aid studies on noise-reduction approaches are presented, that demonstate different use cases.