Objective listening quality assessment of speech communication systems introducing continuously varying delay (time-warping): A time alignment issue
Ludovic Malfait, Phil Gray, Martin J. Reed · IEEE International Conference on Acoustics Speech and Signal Processing · 2008
Standardized in 2001, ITU-T Rec. P.862 (PESQ) is the current "in-force" ITU-T standard for intrusive listening speech quality assessment and is widely used in the speech communication industry. It has proved to be reliable in the assessment of the quality of signals transmitted through traditional communication networks (landline, mobile and first generation of voice over IP systems). However, with the development of internet telephony, like Skype or MSN, advanced techniques for packet concealment and jitter buffer adaptation maybe used, introducing continuously varying delay and making PESQ unreliable. This paper finds that the alignment resolution required by the next generation PESQ-like algorithms to accurately predict subjective tests that evaluate systems that incorporate time-warping would be plusmn 5 ms.