Measuring audio quality is particularly difficult because the measurement methodology itself strongly biases the results. While a previous paper by the same author covered a broad range of biases, this report focuses only on five types of systemic error potentially affecting quantifying judgments: range equalization bias, stimulus spacing bias, contradiction bias, and biases due to nonlinear properties of the assessment scale. These biases are prevalent in audio and speech quality evaluations. Empirical data obtained by various researchers over the past fifteen years was used to illustrate biases in a graphical representation. The results conclusively show that assessment methods are inherently relative. These results also raise important questions about the utility of verbal descriptors. Researchers should avoid conclusions about quality by associating numerical scores with verbal descriptors at fixed positions along the scale.
Affiliation: Faculty of Computer Science, Bialystok University of Technology, Poland
JAES Volume 64 Issue 1/2 pp. 55-74; January 2016
Publication Date: February 5, 2016
Download Now (460 KB)
No AES members have commented on this paper yet.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.