Effects of NAL-R amplification on consonant speech perception in hearing-impaired listeners

Hearing Impairment;Consonant-Vowel Perception;Hearing Aids

Scheidiger, Christoph ; Allen ; Jont B.

关键词: Hearing Impairment;    Consonant-Vowel Perception;    Hearing Aids;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/46572/Christoph_Scheidiger.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This thesis investigates speech perception in hearing impaired (HI ) subjects. Psychoacoustic experiments in different conditions were undertaken. In particular, two consonant vowel (CV ) identification experiments in masking noise were conducted at various signal-to-noise ratios (SNRs) with 16 HI ears. In one of the experiments, the CVs were presented with a uniform gain; in the other experiment, a spectral compensation (i.e. NAL–R) for the individual hearing loss was provided. In both gain conditions, the subjects were instructed to adjust the presentation level to their most comfortable loudness (MCL), which is contrary to the common approach of adjusting the presentation level depending on the pure tone thresholds (PTTs) and the long-term average speech spectrum (LTASS ) (Zurek and Delhorne (1987), Posner and Ventry (1977)). The data demonstrated that the MCL approach led to consistent responses in all subjects. Based on these results, a more rigorous definition of audibility based on entropy and the Miller and Nicely (1955) confusion groups is proposed. Furthermore, the effectiveness of NAL–R for CV perception was investigated by comparing the confusion matrices of the two experiments. In general, the error and entropy decreased with NAL–R. The average error decreased from 20.1% (σ = 3.7) to 16.3% (σ = 2.8). It was also shown that, with the help of NAL–R, the tested ears became more consistent in their responses for a given token. However, for 15.1% of the token-ear pairs (TEPs), the entropy and error increased with NAL–R. It was shown that these 15.1% of the TEPs contained all ears and a large variety of tokens. A method based on the Hellinger Distance (HD) was introduced that enabled comparison of rows of confusion matrices and to calculate distances between responses. With this method, the highly individual problems of the 15.1% of the TEPs were further investigated and compared to the results obtained in normal hearing subjects. In conclusion, it is argued that speech testing — using the proposed methods and experiments as described in this thesis — can deliver valuable and reliable information about individual hearing loss that goes beyond what can be achieved using pure tone thresholds.