Enhanced intrusive Voice Quality Estimation (EVQE)

A well known and used method of intrusive objective quality assessment is ITU-T P.862 — “Perceptual Evaluation of Speech Quality (PESQ) suffers from certain weaknesses. The standard is falling behind the latest coding technologies and as some papers have shown lacks the sufficient accuracy in performing reliable and accurate estimation. The proposed method of Enhanced Voice Quality Estimation (EVQE) utilizes a combination of several effective approaches from the field of speech signal analysis and voice quality estimation. Each approach presents its own perspective to assemble a comprehensive model of the human auditory system and perception. We found a considerable improvement in performance both in terms of the correlation and the absolute estimation error, by adapting these perceptual models to fit the requirements of intrusive speech quality evaluation algorithm, and combining them with PESQ.