Conventional objective means of characterizing communications systems are often inadequate. In particular when a system contains nonlinear elements, such as low-bit-rate encoders, some types of speech transducer, active gain control, and echo cancelers, simple steady-state measurements are not sufficient to predict subjective performance. The use of speechlike test stimuli to characterize nonlinear systems is not new, but a new speechlike composite signal is presented which works in conjunction with a new perception-based objective measure to provide a versatile measurement technique. Rigorous subjective validation of the new method has still to be completed, but early results are given which indicate the usefulness of the method to provide a perceptual ranking for communication systems and also as a diagnostic tool.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.