Inter-rater reliability of accessing the intelligibility of band-limited transformed speech using nonlinear frequency compression

Speech intelligibility verification through a listening test session by human raters was conducted to assess the intelligibility of band-limited speech. The band-limited speech was generated using nonlinear frequency compression via spectral envelope transformation to assist individuals with high-frequency hearing loss. However, addressing the issue on the reliability of the assessment method is important because of the human factors that cause variability in the intelligibility test results. This study investigated the use of Cohen´s Kappa coefficient to measure the inter-rater reliability in the intelligibility test. The inter-rater reliability was found to be high when Cohen´s Kappa coefficient was over 0.6. The result indicates that the human raters showed a high level of agreement in assessing the intelligibility of the band-limited speech.