LOUDNESS MATCHING OF SIGNALS SPECTRALLY SHAPED BY A SIMULATED HEARING-AID

A hearing aid with multiple frequency responses was simulated by programming an equalizer to produce spectral tilt factors of -6, 0, and +6 dB/octave over the frequency range from 0.25 kHz to 4 kHz. Listeners with normal hearing matched the loudness of signals (speech and white noise) that were shaped by these different equalizer settings and delivered via an insert earphone. All signals with spectra that were tilted, either negatively or positively, were perceived as louder than untilted signals. The general pattern of loudness matching was similar across subjects, and intrasubject judgments were found to be highly transitive. A measure of signal power was found to account only moderately well for the individual data. Preliminary evidence from a follow-up study using tilt factors less severe than those used in the main experiment suggests that loudness differences are roughly proportional to the degree of spectral tilt. The incorporation of level corrections approximating those necessary to achieve equal loudness is recommended in the fitting of programmable hearing aids.