Article
Investigation of speech recognition and listening effort in acoustically complex scenes
Search Medline for
Authors
-
Theresa Jansen
- Hörzentrum Oldenburg gGmbH, Exzellenzcluster „Hearing4all“, Oldenburg, DE
-
Sophie Jäker
- Hörzentrum Oldenburg gGmbH, Institut für Hörtechnik und Audiologie, Jade Hochschule, Oldenburg, DE
-
Laura Hartog
- Hörzentrum Oldenburg gGmbH, Exzellenzcluster „Hearing4all“, Oldenburg, DE
-
Melanie Krüger
- Hörzentrum Oldenburg gGmbH, Exzellenzcluster „Hearing4all“, Oldenburg, DE
-
Inga Holube
- Institut für Hörtechnik und Audiologie, Jade Hochschule, Oldenburg, DE
-
Hendrik Kayser
- Hörzentrum Oldenburg gGmbH, Exzellenzcluster „Hearing4all“, Oldenburg, DE
| Published: | March 1, 2023 |
|---|
Outline
Text
Speech recognition performance is measured to diagnose a hearing impairment or to quantify the benefit of hearing devices. Typically, such measurements are conducted in acoustic scenarios with simple sound source configurations, i.e., presentation of a target speech signal from the front and a stationary noise source from the same direction (S0N0) or one lateral position (e.g., S0N90). However, these laboratory measurement conditions do not reflect real-life listening environments and communication situations relevant for people with hearing impairment. For instance, aspects of spatial listening, such as binaural unmasking and better-ear listening are either not addressed (S0N0) or overly prominent (S0N90) compared to realistic scenarios. Furthermore, in a typical clinical setting, only speech recognition is considered, which covers only one aspect of speech perception. Another measure that provides information about speech perception is listening effort which captures aspects of communication even if no decrease of speech recognition is measured. Therefore, the aim of this study was to design acoustically complex scenes in which listening effort as well as speech recognition can be reliably assessed. In the composition of the scenes, different effects were considered, namely spatial release from masking, dip listening, reverberation, energetic and informational masking. Measurements were conducted with 15 normal-hearing and 5 hearing-impaired participants. Speech recognition thresholds were measured with the German matrix sentence test [1]. Listening effort was obtained with the ACALES procedure [2]. All participants passed the measurement twice with a time gap of at least one week. During each appointment 10 complex scenes were presented in random order. They differed in the spatial configuration of the sources, noise type, and number of sources. The virtual acoustic scenes were rendered with the Toolbox for Acoustic Scene Creation and Rendering (TASCAR) [3] and were presented via 16 loudspeakers in a sound-proof listening booth. Results for the test-retest reliability in the scenes will be presented. The different scenes will be compared and the applicability of the scenes for future measurements will be assessed.
References
- 1.
- Wagener K, Brand T, Kollmeier B. Development and evaluation of a German sentence test Part II: Optimization of the Oldenburg sentence test. Zeitschrift für Audiologie. 1999;38:44–56.
- 2.
- Krueger M, Schulte M, Brand T, Holube I. Development of an adaptive scaling method for subjective listening effort. The Journal of the Acoustical Society of America. 2017; 141(6):4680–93. DOI: 10.1121/1.4986938
- 3.
- Grimm G, Luberadzka J, Hohmann V. A toolbox for rendering virtual acoustic environments in the context of audiology. Acta Acustica united with Acustica. 2019;105(3):566–78. DOI: 10.3813/AAA.919337
