Self-Fitting Algorithm for Digital Hearing Aid Based on Interactive Evolutionary Computation and Expert System

Abstract

The traditional hearing aid fitting method, which mainly relies on the audiologist, is timeconsuming and messy. To improve this situation, a self-fitting algorithm based on an improved interactive evolutionary computation (IEC) algorithm and expert system, which enables the patients to fit the hearing aid by themselves, is proposed. The algorithm takes the band gain as the fitting target and uses the patient’s subjective evaluation to iteratively update the algorithm parameters based on the improved IEC algorithm. In addition, a real-time updated expert system is constructed to assist in the optimization of the initial and iterative parameters of the fitting based on the patient’s audiogram and personal information. To verify the performance of the algorithm, a self-fitting software for the hearing aid is designed. Through this software, the test signal is generated for the patient to evaluate the audio quality on a five-level scale. Based on the evaluation results, the algorithm iteratively optimizes the algorithm parameters until the patient is satisfied with the generated audio. Compared with the fitting algorithm based on Gaussian processes algorithm or the interactive evolutionary algorithm, the average subjective speech recognition rate of the proposed algorithm increase at least 11%. The average recognition rate for environmental sound is also improved by at least 2.9%. In addition, the fitting time of the proposed algorithm is shortened by at least 10 min compared to others two algorithms.

Keywords

• hearing aid
• self-fitting
• interactive evolutionary computation
• expert system