Frequency‐specific data from the FAAF speech identification test

Abstract

Can the known selective effects of electronic or physiological distortions upon reception of speech features be used to infer the distortions suffered? At Salt Lake City we described the FAAF test and methodology. By classifying patterns of errors in full symmetrical sets from reference experiments with band‐pass filtering in quiet, subscores were defined to quantify use of speech information from low‐, mid‐, and high‐frequency bands. Now we have related subscores to psychoacoustical tuning curve width and to temporal aspects of processing in normal and sensorineural impaired listeners. The correlation of a composite of six high‐frequency psychoacoustical measures with the high‐frequency band subscore was 0.84, with the low‐band subscore 0.53, confirming the frequency specificity of the subscores. Although FAAF subscores might be less efficient than total scores on the specialised tests, the subscore approach combines a usable degree of specificity on several dimenions with the low administrative cost of a single general‐purpose test. Applications include diagnostic assessment of mild‐moderate impairments and high‐quality communication channels.