Evaluating hearing aid amplification using idiosyncratic consonant errors.

The goal of this study is to provide a metric for evaluating a given hearing-aid insertion gain using a consonant recognition based measure. The basic question addressed is how treatment impacts phone recognition at the token level, relative to a flat insertion gain, at the most-comfortable-level (MCL). These tests are directed at fine-tuning a treatment, with the ultimate goal of improving speech perception, and to identify when a hearing level gain-based treatment degrades phone recognition. Eight subjects with hearing loss were tested under two conditions: flat-gain and a treatment insertion gain, based on subject's hearing level. The speech corpus consisted of consonant-vowel tokens at different signal to speech-weighted noise conditions, presented at the subject's MCL. The treatment caused the average score to improve for 31% of the trials and decrease for 12%. An analysis method based on the accumulated error differences was devised to quantify the benefit each individual ear received from the treatment. Using this measure, the effect of the treatment could be evaluated, providing precise characterization of idiosyncratic phone recognition. This analysis directs the audiologist toward the most susceptible subject-dependent tokens, to focus in the process of fine-tuning the insertion gain of the hearing-aid.