Various articles suggest that better speech understanding can be obtained by auditory training. Auditory training typically consists of training speech perception in varying background noise levels or with degraded speech signals. Recently a Danish material for auditory training has been developed. This material consists of music training examples as well as speech training exercises. The rationale behind adding music training is variation in exercises as well as calling attention to details in auditory perception that could be valuable in speech perception as well as in hearing aid fitting. The results presented in this poster originate from examination of the benefits this material can provide on speech perception. Results from the investigation show an average benefit of auditory training, but with a large interpersonal variation, suggesting that a preselection of the individuals better suited for auditory training is needed. A battery of cognitive tests has been applied pre- and post-training, results from these tests are presented and discussed, in order to determine if there is correlation between cognition in general, improvement in cognition by auditory training, and obtaining better speech understanding by auditory training.
Some of the challenges that hearing-aid listeners experience with speech perception in complex acoustic environments may originate from limitations in the temporal processing of sounds. To systematically investigate the influence of hearing impairment and hearing-aid signal processing on temporal processing, temporal modulation transfer functions (TMTFs) and “supra-threshold” modulation-depth discrimination (MDD) thresholds were obtained in normal-hearing (NH) and hearing-impaired (HI) listeners with and without wide-dynamic range compression (WDRC). The TMTFs were obtained using tonal carriers of 1 and 5 kHz and modulation frequencies from 8 to 256 Hz. MDD thresholds were obtained using a reference modulation depth of −15 dB. A compression ratio of 2:1 was chosen. The attack and release time constants were 10 and 60 ms, respectively. For both carrier frequencies the TMTF thresholds decreased with the physical com-pression of the modulation depth due to the WDRC. Indications of reduced temporal resolution in the HI listeners were observed in the TMTF patterns for the 5-kHz carrier. Significantly higher MDD thresholds were found for the HI group relative to the NH group. No relationship was found between the MDD thresholds and the TMTF threshold. These findings indicate that the two measures may represent different aspects of temporal processing.
Modern digital hearing aids require and offer a great level of personalization. Today, this personalization is not performed based directly on what the user actually perceives, but on a hearing-care professional’s interpretation of what the user explains about what is perceived. In this paper, an interactive personalization system based on Gaussian process regression and active learning is proposed, which personalize the hearing aids based directly on what the user perceives. Preliminary results demonstrate a significant difference between a truly personalized setting obtained with the proposed system and a setting obtained by the current practice.
