No abstract

We investigated the performance of the objective speech quality assessment method called "wideband PESQ", which has been proposed to ITU-T for quality assessment of wideband speech, for quality ranging from narrowband to wideband speech with packet-loss degradation. The experimental results show that although the relationship between wideband PESQ and subjective quality depends on the codec for wideband speech, wideband PESQ can be used quite well as an objective quality measurement for speech with packet-loss degradation as long as the same codec is used. This codec dependence, which is especially observed when evaluating wideband codecs, makes it difficult to consistently estimate the quality of speech coded by various kinds of codecs. To avoid this difficulty, we propose correcting objective scores based on a function prepared for each codec. This correcting function is predetermined based on the results of a subjective quality experiment. We show the validity of the proposed method in evaluating the speech quality of various codecs with packet-loss degradation.

No abstract

No abstract

No abstract