Institute for Telecommunication Sciences / Research Topics / Audio Quality Research / Other Work and Info / Quality Estimation
Objective Estimation of Perceived Audio or Speech Quality
Objective estimators of perceived audio or speech quality seek to provide meaningful, repeatable, rapid estimates of perceived quality. The most fundamental and correct measures of speech quality come from listening experiments or subjective tests. However, carefully conducted listening experiments tend to be fairly complex and time consuming, and the associated costs make them inappropriate for some applications. Thus, objective estimators may offer practical alternatives to listening experiments in some situations.
Objective estimators typically apply digital signal processing algorithms to "sent" and "received" to estimate the difference between them in a perceptually consistent way. This difference can then be mapped to an estimate of perceived speech quality, often using one of the scales that are used in listening experiments. The advantages of objective estimators over listening experiments are clear. The disadvantage is also very clear: objective estimators can only provide estimates. Some of the technical publications found on this site describe Program efforts in the area of quality estimation.
For objective estimation of speech quality one can consider
- POLQA: ITU-T Rec. P.863 (01/2011), "Perceptual Objective Listening Quality Assessment"
- PESQ: ITU-T Rec. P.862 (02/2001), "Perceptual Evaluation of Speech Quality"
- MNB: ATIS-0100518.1998 (R2013), "Objective Measurement of Telephone Band Speech Quality Using Measuring Normalizing Blocks (MNBs)"
For estimation of 20 kHz bandwidth music quality one can consider
- PEAQ: ITU-R Rec. BS-1387 " Method for Objective Measurements of Perceived Audio Quality"