Comparing recognition of distorted speech using an equivalent signal-to-noise ratio index

An index of equivalent performance in noise was developed to compare recognition in different forms of speech distortion. Speech-recognition performance of young and elderly listeners with and without hearing loss was evaluated for undistorted speech presented in quiet and noise, and for speech distorted by four time-compression ratios and by four reverberation times. The data obtained in noise on young subjects with normal hearing served to generate a normalized regression equation, which was used to convert percent-correct performance in different distortion conditions to equivalent performance for undistorted speech at a particular S/N ratio. Comparisons of the equivalent S/N ratios obtained in the various conditions allowed rank-ordering of speech recognition performance in different types of degradation. The data also show that age and hearing loss affect recognition of speech degraded by reverberation or time compression. However, age effects are evident primarily in the most severe distortion conditions. Recognition of undistorted speech in noise was affected by hearing loss but not by age. These findings support a hypothesis that stipulates that increased age produces a reduction in the functional S/N ratio.