Speech recognition and loudness perception in normal-hearing and hearing-impaired listeners
Year: 2020
Type: conference-paper
Abstract: Understanding the individual listening performance of hearing-impaired listeners is crucial for improving auditory rehabilitation strategies and for user satisfaction. Here, speech recognition of normal-hearing and hearing-impaired listeners for well controlled laboratory conditions as well as for realistic auditory scenes are considered. For hearing-impaired listeners two hearing aid prescription rules are compared: NAL-NL2, which is commonly used in commercial hearing devices, and trueLOUDNESS (Oetting et al., 2018). While NAL-NL2 fitting is based on the pure tone threshold, the trueLOUDNESS uses loudness measurements for gain determination. The prescription rules are compared in terms of speech recognition and loudness perception. Speech reception thresholds (SRTs) were measured in four noise conditions: stationary, speech-shaped noise, modulated ICRA5-250 noise, and two realistic scenes, nature, and cafeteria (Grimm et al., 2016). Loudness ratings were determined for both scenes and real-world test signals to investigate the loudness perception. Hearing aid fitting was provided via the Master Hearing Aid (Grimm et al., 2006). The realistic auditory scenes were implemented in TASCAR (Grimm et al., 2019). trueLOUDNESS provided better SRTs than NAL-NL2 for low levels of background noise. At higher levels of 75 dB comparable results were observed. In all aided conditions, the hearing-impaired listeners performed worse than normal-hearing listeners. The SRT benefits in the standard laboratory conditions (stationary and modulated noise) correlated well with the SRTs in realistic scenes but only for a low level of background noise. Binaural loudness perception for broadband signals was close-to-normal with trueLOUDNESS lowe-than-normal with NAL-NL2. In order to better understand the contribution of hearing threshold and suprathreshold deficits on speech recognition, the predictions of a binaural speech intelligibility model will be presented as well.
Source: HAL
Authors: Kramer, Florian, Oetting, Dirk, Schädler, Marc René, Hohmann, Volker, Warzybok, Anna
Institution: Carl von Ossietzky Universität Oldenburg
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W4287556390