Timbre perception and object separation with normal and impaired hearing

Timbre is a combination of all auditory object attributes other than pitch, loudness and duration, and is used to distinguish different musical instruments or voices. People with sensorineural hearing loss often have problems with timbre distortion. Even for modern hearing aids it is difficult to provide good audio quality for speech intelligibility while preserving the natural timbre. This not only affects music perception, but may also influence object recognition in general. The present study aims to quantify differences in object segregation and timbre discrimination between normal-hearing and hearing-impaired listeners with a sensorineural hearing loss. In order to improve auditory models and hearing aids, a new method for studying timbre perception was developed. Using cross-faded (morphed) instrument sounds in psychoacoustic measurements, the subtle timbre perception differences between listener groups are studied. In order to characterize timbre perception differences, rating measurements were performed, in which normal-hearing and hearing-impaired subjects judged the similarity of the presented morphed sounds. When stimuli were amplified to provide intermediate loudness impressions in all subjects, most hearing-impaired subjects gave ratings similar to those of normal-hearing listeners. Only a few subjects showed distinct rating deviations from normal-hearing listeners. In order to verify subtle perception differences, the morphed stimuli were combined with discrimination measurements. Experiments with normal-hearing musicians and non-musicians showed that the new method enables objective determination of a value that provides a comparison between different subject groups and timbres: a just noticeable difference (JND) of timbre. For the discrimination measurements, the attack portion of the sound was cut off, which minimizes recognition of the sounds and thus makes the method independent of subjects’ previous knowledge. Discrimination measurements with normal-hearing and hearing-impaired listeners aim to quantify differences in object segregation and timbre discrimination, investigating timbre JNDs in silence and different background-noise conditions, on different sound levels and in subjects with different hearing loss configurations. The results indicate that at intermediate levels JNDs of subjects with flat or diagonal hearing loss are similar to those of normal-hearing listeners, when an appropriate linear sound amplification is provided. This contradicts the common hypothesis that hearing-impaired people generally have more problems in distinguishing different timbres, for example, due to reduced frequency selectivity. However, subjects with a steep hearing loss show significantly higher JNDs than normal-hearing listeners, both in silence and in noise. In the condition testing transferability from silence to