Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Electrical and Computer Engineering


Vijay Parsa


Modern digital hearing aids provide an array of features to improve the user listening experience. As the features become more advanced and interdependent, it becomes increasingly necessary to develop accurate and cost-effective methods to evaluate their performance. Subjective experiments are an accurate method to determine hearing aid performance but they come with a high monetary and time cost. Four studies that develop and evaluate electroacoustic hearing aid feature evaluation techniques are presented. The first study applies a recent speech quality metric to two bilateral wireless hearing aids with various features enabled in a variety of environmental conditions. The study shows that accurate speech quality predictions are made with a reduced version of the original metric, and that a portion of the original metric does not perform well when applied to a novel subjective speech quality rating database. The second study presents a reference free (non-intrusive) electroacoustic speech quality metric developed specifically for hearing aid applications and compares its performance to a recent intrusive metric. The non-intrusive metric offers the advantage of eliminating the need for a shaped reference signal and can be used in real time applications but requires a sacrifice in prediction accuracy. The third study investigates the digital noise reduction performance of seven recent hearing aid models. An electroacoustic measurement system is presented that allows the noise and speech signals to be separated from hearing aid recordings. It is shown how this can be used to investigate digital noise reduction performance through the application of speech quality and speech intelligibility measures. It is also shown how the system can be used to quantify digital noise reduction attack times. The fourth study presents a turntable-based system to investigate hearing aid directionality performance. Two methods to extract the signal of interest are described. Polar plots are presented for a number of hearing aid models from recordings generated in both the free-field and from a head-and-torso simulator. It is expected that the proposed electroacoustic techniques will assist Audiologists and hearing researchers in choosing, benchmarking, and fine-tuning hearing aid features.