Relation Between Different Measures of Feedback in Hearing Aids and Gain Measures Derived from Percentile Analysis

Feedback management is a common feature of today's hearing aids. A huge variety of algorithms exist with the objective to control feedback effectively while achieving a high added stable gain, good sound quality and with a fast adaptation to changes in feedback paths. Because the real-life performance of a hearing aid is significantly affected by its feedback behavior, objective procedures for characterizing feedback management systems are desirable. This study investigates the possibilities of characterizing feedback management systems of commercial hearing aids by estimating the insertion gain derived from percentile analysis (IEC draft standard 60118-15 [1]) and analyzing its relation to existing direct measures of feedback. Responses of 10 hearing aids from 5 manufacturers with different signal processing and different feedback control algorithms were recorded in static conditions, using a dummy head setup and realistic test signals. Several instrumental measures of provided insertion gain and distortion related to feedback and its control were calculated from the recordings. Results show that the insertion gain is largely different across devices at the same hearing loss setting, which leads to pronounced differences in feedback susceptibility. Furthermore, the difference in estimated insertion gains for the same device operating with and without feedback control algorithm was found to be indicative of feedback-related problems. High correlations were found between feedback measures from insertion gain data and direct feedback measures, indicating that percentile analysis is applicable in principle to characterizing feedback behavior. The relation of instrumental measures with perceptual performance measures, however, remains to be established.