We can better estimate an audio system if errors can be assigned audible significance. Recently established data of the auditory filter is developed into models that predict the audibility, detectability and loudness or errors. Errors are assessed by frequency, loudness, temporal and spatial summations. The models explain critical-band phenomena and several types of masking anomalies. Particular attention is paid to determining the audibility of different noise-bands, especially near absolute and masked thresholds, with experimental verification. Worked examples include dither noise and distortion. Realized as computer post-processors for conventional instruments, the approach emphasizes graphical and numerical output.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.