Perceptual Evaluation of Speech Quality

Perceptual Evaluation of Speech Quality, known as PESQ, is a family of standards that gives a test methodology for the automated assessment of speech quality as a user of a telephony system experiences it. It was standardized as Recommendation ITU-T P.862 in 2001 and is used for objective voice quality testing by phone manufacturers, network equipment vendors and telecom operators, although using it requires a license. PESQ was built to model the subjective listening tests of telecommunications, so it takes real voice samples as test signals rather than tones or noise. It is a full-reference algorithm: it compares the degraded signal at the listener side with the original reference signal sample by sample after a temporal alignment of corresponding excerpts, and its results principally model mean opinion scores on a scale from 1 (bad) to 5 (excellent). Related recommendations map PESQ scores to a MOS scale (P.862.1), cover wideband measurements (P.862.2) and give an application guide (P.862.3). It superseded the 1997 PSQM recommendation P.861, and its successor POLQA (P.863) entered into force in 2011.