Intelligibility Estimation in Law Enforcement Speech Processing
Patrick A. Naylor, Nikolay D. Gaubitch, D. Sharma, Gaston Hilkhuysen, Mark A. Huckvale · 2010
Speech recordings obtained in the context of law enforce-ment are often degraded in terms of quality and intelligib-ility. Several techniques for assessing the impact of speech enhancement algorithms on quality are available, both in-trusive and nonintrusive, but the assessment of intelligibil-ity is usually reliant on expensive and time consuming sub-jective listening scores. To address this issue, we describe some recent scoring experiments and an adaptive Bayesian procedure which efficiently estimates properties of the psy-chometric function from a small number of listening tests. A data-driven nonintrusive objective intelligibility estima-tion method is also described and tested on car and babble noise. It is shown to give intelligibility estimates that are well correlated with subjective scores. We aim to im-prove our understanding of the quality/intelligibility trade-off and to study speech processing tools in the critical con-text of law enforcement. 1