Related Experiment Videos
Measures of hearing aid gain for real speech
P G Stelmachowicz1, J Kopun, A L Mace
1Boys Town National Research Hospital, Omaha, Nebraska, USA.
Objective:
Non-speech stimuli typically are used to estimate the electroacoustic characteristics of a hearing aid. At present, there is no consensus as to what type of input stimulus will best represent the gain for real speech. The purpose of this study was to measure hearing aid gain using continuous discourse and to compare these values with gain measured with five different types of simpler stimuli.
Design:
Hearing aid gain as a function of frequency was measured in a 2 cm3 coupler for 20 commercially available hearing aids. Circuitry included features such as linear peak clipping, compression limiting, 1-, 2-, and 3-channel full dynamic range compression, and adaptive compression. Input stimuli were a) 15 sec of continuous discourse, b) swept pure tones (SPTs), c) speech weighted composite noise (SWCN), d) simulated speech, e) speech weighted warble tones, and f) speech modulated noise. Input levels ranged from 50 to 80 dB SPL.
Results:
In general, both SPTs and SWCN tended to underestimate the high-frequency gain for real speech. These discrepancies increased as a function of input intensity. On average, the SPT produced the greatest departure from the gain for real speech, producing differences for individual hearing aids as large as 10 to 14 dB. An analysis by circuit type revealed that discrepancies most likely occurred when a hearing aid was operating in a nonlinear mode. Of the five non-speech stimuli used, speech modulated noise and simulated speech seemed to provide the closest approximation to the gain measured with continuous discourse.
Conclusions:
When a hearing aid is operating in a nonlinear mode, non-speech stimuli will tend to underestimate the gain for real speech, particularly in the high frequencies. Under some conditions, these discrepancies may impact clinical decisions during the hearing aid fitting process. Additional studies are needed to elucidate the factors that contribute to the gain discrepancies observed in this study and to explore the use of additional stimuli (including short speech samples), which may result in better predictions of the gain for speech.