Spatial audio quality perception (part 2): a linear regression model

The QESTRAL (Quality Evaluation of Spatial Transmission and Reproduction using an Artificial Listener) system is intended to be an artificial-listener-based evaluation system capable of predicting the perceived spatial quality degradations resulting from SAPs (Spatial Audio Processes) commonly encountered in consumer audio reproduction. A generalizable model employing just five metrics and two principal components performs well in its prediction of the quality over a range of program types. Commonly-encountered SAPs can have a large deleterious effect on several spatial attributes including source location, envelopment, coverage angle, ensemble width, and spaciousness. They can also impact timbre, and changes to timbre can then influence spatial perception. Previously obtained data was used to build a regression model of perceived spatial audio quality in terms of spatial and timbral metrics. In conjunction with two simple probe signals, the resulting model can form the core of an evaluation system.