Measuring and Predicting Perceived Audio Quality
[Feature Article] It is widely accepted that listening tests are time-consuming and expensive to conduct, yet they remain the only reliable way of evaluating perceived sound quality in a number of fields. The Holy Grail of psychoacoustics research, at least in many quarters, is the ability to model the quality perception process accurately enough so that measurements of the audio signal can be converted into predictions of perceived quality. However, this is a complex issue. The judgment of sound quality depends on many perceptual and cognitive factors. In tests to judge speech quality an objective measurement has been possible for some time now, with various measures in existence designed to predict the quality of speech coded at low bit rates. More recently there have been standard models to predict the quality of full-range audio signals, and again these have tended to be driven by the introduction of perceptual audio coding at low bit rates. The accuracy of prediction varies, and researchers aim to improve this with refinements to the models they use.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.