Abstract
Many researchers have proposed that multi-channel compression hearing aids should process sounds so as to restore loudness perception to ‘normal’. However, procedures for achieving this have generally been based on measurements or calculations using narrowband stimuli, and these procedures may not be accurate for broadband sounds such as speech. Here, a model for predicting loudness for people with cochlear hearing loss is used to calculate the frequency- and level-dependent gains that would be required to restore loudness perception to ‘normal’ for speech-like signals. The calculations are based entirely on the pure tone audiogram, and do not require measures of loudness growth. The model was applied to several different hypothetical hearing losses, varying in slope and severity. In each case, the model was used to calculate the insertion gains (IGs) that would be required as a function of frequency so that speech-shaped noise with a level of 65 dB SPL would evoke a specific loudness pattern matching that for a normal ear. A similar procedure was applied using speech-shaped noise with a level of 85 dB SPL (with the spectral characteristics of shouted speech). The results were used to derive functions relating the required IG to hearing loss for each audiometric frequency and each speech-shaped noise level. These functions were used in turn to derive compression ratios and gains for each channel of a multi-channel compression system. The derivations apply to systems with any number of channels. The outcome is a method than can be used for the initial fitting of multichannel compression hearing aids, so as to restore loudness perception to near ‘normal’ for broadband speech-like signals.
