Efficient characterization of individual differences in compression ratio preference

J Acoust Soc Am. 2018 Nov;144(5):2662. doi: 10.1121/1.5067390.

Abstract

While wide dynamic range compression (WDRC) is a standard feature of modern hearing aids, it can be difficult to fit compression settings to individual hearing aid users. The goal of the current study was to develop a practical test to learn the preference of individual listeners for different compression ratio (CR) settings in different listening conditions (speech-in-quiet and speech-in-noise). While it is possible to exhaustively test different CR settings, such methods can take many hours to complete, making them impractical. Bayesian optimization methods were used to find CR preferences in individual listeners in a relatively short amount of time. Using this practical preference learning test, individual differences in CR preference were examined across a relatively wide range of CR settings in different listening conditions. In experiment 1, the accuracy of the preference learning test in normal hearing listeners was verified. In experiment 2, it is shown that individual hearing impaired listeners differ in their CR preferences, and listeners tended to prefer the CR setting identified by the preference learning test over both linear gain or the National Acoustics Lab--Nonlinear 2 CR prescription based on their audiograms.

Publication types

  • Comparative Study

MeSH terms

  • Acoustic Stimulation / methods
  • Adult
  • Aged
  • Algorithms
  • Auditory Perception / physiology*
  • Bayes Theorem
  • Data Compression
  • Female
  • Fourier Analysis
  • Hearing Aids / trends*
  • Hearing Tests / methods
  • Humans
  • Individuality
  • Male
  • Middle Aged
  • Noise / adverse effects*
  • Patient Preference / statistics & numerical data*
  • Persons with Hearing Disabilities / rehabilitation
  • Persons with Hearing Disabilities / statistics & numerical data