Extension of ITU-T Recommendation P.862 PESQ towards Measuring Speech Intelligibility with Vocoders

摘要

ITU-T recommendation P.862 PESQ was developed for assessing speech quality. The basic idea in PESQ is to compare a reference speech signal with the degraded signal through a psycho-acoustic model and a model of human quality comparison (cognitive model). Within NATO, testing of low bit rate speech codecs is more focused on speech intelligibility than on speech quality. Although the cognitive model of PESQ was designed to represent the quality judging process, it was already found that, for specific applications, PESQ can also predict intelligibility. In this paper PESQ is validated for assessing speech intelligibility with low bit rate vocoders. The results show that improvements in PESQ are necessary in order to obtain high correlations between objective and subjective intelligibility scores.