By Topic

Objective Prediction of the Sound Quality of Music Processed by an Adaptive Feedback Canceller

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
Manders, A.J. ; Instn. of Sound & Vibration Res., Southampton Univ., Southampton, UK ; Simpson, D.M. ; Bell, S.L.

Adaptive feedback cancellers in hearing aids can produce unpleasant sounding distortion artifacts (entrainment) in response to periodic inputs, including music. Reliable objective metrics that predict user-perceived distortion could significantly reduce development costs for new hearing aids. The aim of this study was to gain insight into the ability of different objective metrics to predict subjective ratings of the sound quality of music processed by adaptive feedback cancellation. The metrics tested consisted of perceptual measures from established audio quality models (including Perceptual Evaluation of Audio Quality (PEAQ), PEMO-Q and .Rnonlin). Neural networks were used to map between the values of the perceptual measures and a subjective scale of perceived quality. Training data consisted of values of perceptual measures obtained from ten different excerpts of orchestral music processed by a simplified model of a hearing aid with an adaptive feedback canceller, and corresponding subjective quality ratings from 27 normal hearing subjects. An optimal combination of perceptual measures to use as inputs to a network input was found using an extended Fourier amplitude sensitivity test (EFAST). Our results suggest that the most salient inputs to a multivariate model of measured quality ratings consist of perceptual measures related to spectral noise loudness, modulation differences between clean and processed signals, and correlation-based measurement of nonlinear distortion. The intraclass correlation between mean subjective ratings and the output of a network combining these perceptual measures was high , which compares favorably to results from previous studies of perceptual quality metrics applied to audio signals with other forms of noise or distortion.

Published in:

Audio, Speech, and Language Processing, IEEE Transactions on  (Volume:20 ,  Issue: 6 )