Audio Signal Processing and Coding

12.2 SUBJECTIVE QUALITY MEASURES

Although listening tests are often conducted informally, the ITU-R Recommendation BS.1116 [ITUR94b] formally specifies a listening environment and test procedure appropriate for subjective evaluations of the small impairments associated with high-quality audio codecs. The standard procedure calls for grading by expert listeners [Bech92] using the CCIR continuous impairment scale (Figure 12.1-a) [ITUR90] in a double blind, A-B-C triple stimulus hidden reference comparison paradigm. While stimulus A always contains the reference (uncoded) signal, the B and C stimuli contain in random order a repetition of the reference and then the impaired (coded) signal, i.e., either B or C is a hidden reference. After listening to all three, the subject must identify either B or C as the hidden reference, and then grade the impaired stimulus (coded signal) relative to the reference stimulus using the five-category, 41-point “continuous” absolute category rating (ACR) impairment scale shown in the left-hand column of Figure 12.1 (a). From best to worst, the five ACR ranges rate the coding distortion as “imperceptible (5),” “perceptible but not annoying (4.0–4.9),” “slightly annoying (3.0–3.9)”, “annoying (2.0–2.9),” or “very annoying (1.0–1.9).” A default grade of 5.0 is assigned to the stimulus identified by the subject as the hidden reference. A subjective difference grade (SDG) is computed by subtracting the score assigned to the actual hidden reference from ...