Go To Program 72 On Predicting the Difference in Intelligibility Before and After Single-Channel Noise Reduction Cees Taal 1 Richard Hendriks 1 Richard Heusdens 1 Jesper Jensen 2 1 Delft University of Technology, SIPlab, Netherlands 2 Oticon A/S, Denmark In general, single-channel noise-reduction algorithms do not improve the speech intelligibility for normal-hearing listeners. In order to understand this problem, a reliable objective intelligibility measure is of great interest. Such a measure could be used for analysis and/or optimization of noise-reduction algorithms. For this application it is important that the objective measure can correctly predict the difference in intelligibility before and after noise reduction. Typically, existing studies do not evaluate objective measures for this property. Five objective measures (STOI, CSTI, DAU, CSII and FWS) are evaluated in order to let them predict the intelligibility before and after noise reduction. The measures CSTI, DAU, CSII and FWS significantly overestimated the intelligibility of the noise-reduced speech. This was not the case with STOI, which is therefore a new potential candidate for analysis and/or optimization in the field of single-channel noise reduction. View pdf