hrs@houxb.UUCP (H.SILBIGER) (03/29/84)
In a recent posting about the Carver test for differences between digital and analog audio, the statement was made that people can hear differences as small as 0.5 dB. As a generalization, this is not true. It is true that for some frequency range loudness differences of as little as 0.5 dB can be discriminated. However. this is both level and frequency dependent. Also, this is only true if the two sounds are heard in close succession, i.e. a relative difference in loudness. Over longer time periods, say one minute, the increment in loudness must be much larger. This is called an absolute judgment. These relationships become even more complicated for complex sounds. Any good book on hearing or psychoacoustics has tables and graph detailing some of these relationships. Training can improve the ability to discriminate sounds to some extent, but limits always apply, and these limits are what is reported in the literature. When judging the difference between two sets of stimuli such as analog and digital recordings, one can only tell that they are different, not which one is closer to the original. This could onlyh be done by comparing each to the orioginal performance. However, the difference between even the best possible recording and an original may very well be greater than the difference between the two recordings. An alternative would be to takea master tape (which could be either analog or digital), and re-record it by analog and digital procedures. These re-recordings then should be compared with the master. Proper experimental,procedures (double-blind) should be used of course. That way the answer to the question of which is more like the original could be found. Herman R. Silbiger "Certified Golden Ears" houxb!hrs