Subjective Evaluation of Factors Affecting 2-Channel Stereophony

Abstract

In transmitting and reproducing 2-channel stereophonic signals, the original program material may be modified deliberately or unintentionally. A study was made of many of the subjective aspects associated with these modifications. Three of the more important modifying parameters, involving channel separation, were evaluated in a formal test program using 10 critical observers. Musical selections in groups of four were suitably chosen for each parameter and played through a stereo system with and without modification. Questions were asked for detection of a spatial difference, preservation of the spatial resemblance, and listener preference. Composite results indicate that approximately 90% of the observers say that the spatial resemblance is good if (1) full channel separation is maintained out to 8 kc with no separation above this frequency, (2) full channel separation is maintained down to 100 cycles with no separation below this frequency, (3) a channel separation of 16 db is maintained over the whole band. The results for detection and listener preference are presented graphically along with resemblance data both for individual musical selections and selections in groups of four. In tests using white Gaussian noise radiated from the loudspeakers, it was found that in-phase and out-phase noise as well as two uncorrelated noises have approximately the same loudness (for equal powers). In addition, pertinent observations on other subjective aspects are reported.