Personalization of Hearing AID DSLV5 Prescription Amplification in the Field via a Real-Time Smartphone APP
Aoxin Ni, Edward Lobariñas, Nasser Kehtarnavaz · 2023
This paper presents a real-time smartphone app for the purpose of conducting personalization of hearing aid DSLv5 prescription amplification in the field. The developed smartphone app enables hearing healthcare providers and researchers to study the amplification or compression function of hearing aids in realistic audio environments. The personalization approach is based on a previously developed method of maximum likelihood inverse reinforcement learning (MLIRL) which incorporates a user’s hearing preferences in paired audio comparisons. The app consists of a real-time online training module and a real-time testing module. In the training module, an optimal personalized set of DSLv5 gains across five frequency bands, commonly used for audiograms, is obtained based on paired audio comparisons which are carried out in an on-the-fly or online manner in real-world audio environments or in the field. In the testing module, the MLIRL derived personalized set of gains can be compared to the standard DSLv5 set of gains in terms of hearing preference. The audio processing steps taken to achieve this implementation as a real-time smartphone app are presented and the results related to its ease-of-use and real-time operation are reported.