Evaluation of Electroacoustic Test Signals II: Development and Cross-Validation of Correction Factors

OBJECTIVETo develop and cross-validate corrections for improving the match between amplified speech levels and frequency response measurements with hearing aids. DESIGNPreviously published correction approaches were reviewed. Two regression-based corrections and two nonregression corrections were developed from an existing database of hearing aid responses measured with clinically available test signals and speech (). Corrections were evaluated on a second database of digital hearing aid responses for test signals and speech. The second data set was constructed specifically to challenge three hypothesized threats to the robustness of the corrections. RESULTSThe error for each signal (corrected and uncorrected) was calculated. Correction procedures produced a significant improvement in the match between predicted and measured aided levels of speech. Inclusion of compression-related variables provided small but significant improvements. Results generalized to the second data set. CONCLUSIONSCorrection procedures may be applied to improve the match between aided test signal levels and aided levels of speech.