In 2015 23rd European Signal Processing Conference (EUSIPCO), pages 1063-1067, Aug, 2015. Paper doi abstract bibtex
In this paper the development and evaluation of an extended Reverberation Decay Tail (RDT) metric is described. The signal-based metric predicts the perceived impact of reverberation on speech, by identifying and characterising energy decay characteristics in the signal Bark spectrum. In comparison with a previous metric, the new metric is extended to operate on wideband speech and incorporates an improved perceptual model and decay curve detection scheme. Furthermore, contributions of this work include experimental testing and validation of the metric on reverberant speech. The tests conducted show positive correlation with objective measures such as C50 as well as with subjective listening test scores. Potential applications of the measure include use as an evaluation tool for dereverberation research.