An Instrumental Intelligibility Metric Based on Information Theory

Journal Article (2018)
Author(s)

Steven Van Kuyk (Victoria University of Wellington)

W. Bastiaan Kleijn (Victoria University of Wellington, TU Delft - Signal Processing Systems)

Richard C. Hendriks (TU Delft - Signal Processing Systems)

DOI related publication
https://doi.org/10.1109/LSP.2017.2774250 Final published version
More Info
expand_more
Publication Year
2018
Language
English
Issue number
1
Volume number
25
Pages (from-to)
115-119
Downloads counter
186

Abstract

We propose a monaural intrusive instrumental intelligibility metric called SIIB (speech intelligibility in bits). SIIB is an estimate of the amount of information shared between a talker and a listener in bits per second. Unlike existing information theoretic intelligibility metrics, SIIB accounts for talker variability and statistical dependencies between time-frequency units. Our evaluation shows that relative to state-of-the-art intelligibility metrics, SIIB is highly correlated with the intelligibility of speech that has been degraded by noise and processed by speech enhancement algorithms.