Abstract
11 hearing-impaired (HI) and 12 normal-hearing (NH) subjects have performed sound quality ratings on 6 perceptual scales (Loudness, Clarity, Sharpness, Fullness, Spaciousness and Overall judgement). The signals for the rating experiment consisted of running speech and music with or without background noise. These signals were processed in various configurations of filtering, clipping, and compression to form a total of 64 stimuli. Each stimulus was presented monaurally over headphones and rated three times during successive visits. The stimuli for HI listeners were amplified using the POGO fitting rule.
One major purpose of the experiment was to provide data for an objective measure of sound quality. The obtained data covered a large range on the rating scales, which represented different underlying perceptual scales.
All subjects performed the rating task in a satisfactory manner, but the normal-hearing group was slightly more reliable. There were significant differences between stimuli and between subjects, with stimuli affecting the ratings the most. Normal-hearing and hearing-impaired subjects showed similar trends, but normal-hearing listeners were generally more sensitive, i.e. covered a larger range on each rating scale. Of the chosen signal processing parameters, spectral modifications affected the perceived sound quality the most, but clipping and compression also produced detectable differences.
The perceived sound quality could be described by four underlying perceptual dimensions or, with simpler interpretation, by four of the original rating scales.
The two subject groups agreed in their interpretation of the rating scales, and were almost identical in their use of the scales. Based on this, the rating scales were not considered absolute scales.
One major purpose of the experiment was to provide data for an objective measure of sound quality. The obtained data covered a large range on the rating scales, which represented different underlying perceptual scales.
All subjects performed the rating task in a satisfactory manner, but the normal-hearing group was slightly more reliable. There were significant differences between stimuli and between subjects, with stimuli affecting the ratings the most. Normal-hearing and hearing-impaired subjects showed similar trends, but normal-hearing listeners were generally more sensitive, i.e. covered a larger range on each rating scale. Of the chosen signal processing parameters, spectral modifications affected the perceived sound quality the most, but clipping and compression also produced detectable differences.
The perceived sound quality could be described by four underlying perceptual dimensions or, with simpler interpretation, by four of the original rating scales.
The two subject groups agreed in their interpretation of the rating scales, and were almost identical in their use of the scales. Based on this, the rating scales were not considered absolute scales.
| Original language | English |
|---|
| Volume | Report 51 |
|---|---|
| Number of pages | 133 |
| Publication status | Published - 1992 |