joe@cornell.UUCP (Joseph Bates) (05/06/85)
CD players with sharp anti-aliasing filters exhibit a frequency dependent phase shift of hundreds of degrees at high frequencies. This has been suggested as a cause of harshness in their reproduction. The argument below raises doubt as to the significance of such phase shifts. I would like to learn where (if) the argument fails. Consider a conventional speaker with the tweeter roughly one foot above the woofer. Suppose one is sitting with the woofer at ear level and that in this situation the music is being reproduced "correctly", that is the signals possess the proper phase relationship. Suppose further that one is roughly seven feet from the speaker. [The details here are not critical.] Suppose now that one raises one's head so that it is at the level of the tweeter. Sound from the tweeter reaches the ear somewhat sooner than before (since the tweeter is a bit closer), while sound from the woofer arrives a little later. By my calculations the tweeter is about an inch closer, the woofer an inch further. The cumulative change of two inches is about the length of one sound wave at 6KHz. At 12Khz it is two waves, at 18KHz it is three waves. Thus, the improper positioning of one's head (not its motion, but its improper positioning) appears to introduce a frequency dependent phase shift of many hundreds of degrees at high frequencies. This argument suggest that the vertical positioning of one's head is critical to within inches to avoid a frequency dependent phase shift from the "correct" signal, for otherwise harshness will result. I have not heard of such a problem. I await enlightenment. joe bates (joe@cornell.arpa)
bhs@siemens.UUCP (05/07/85)
I would believe that the problem is a little more complex than that. There are, I believe, two things to consider, neither of which will be affected by speaker/head placements. 1. Analog filters tend to exhibit ringing while settling out. This should not be readily audible while reproducing a low frequency sine wave. However, when trying to transmit a sudden pulse, such as a drumbeat, this ringing could bedome audible. Since the digital filtering processes permit a simpler rear-end filter, this ringing is easier to manage, and thus typically not nearly as pronounced. 2. Speaking of complex sounds, the second point I would like to bring up concerns the Fourier transforms of these sounds, and how the player plays them. If we imagine a relatively complex sound, such as a drumbeat, or a violin note, this sound is composed of various harmonics of the fundamental frequency, where each harmonic sounds at a certain phase. The ultimate examples would be square waves, or triangular waves. The square wave would be square only if the various harmonics are in fact ringing at the neccessary phases, otherwise it would be an almost-square wave. Now, since my theoretical physics has always been my downfall, I will not venture so far as to saying that this aspect will really make a difference audibly. Theoretically, however, it will, and once again, the digital filtering wins, hands down, since it does not introduce any frequency dependent phase shifts. Ultimately, what really counts are mainly subjective variables, as well as your listening environment. Bernard H. Schwab Siemens RTL, Princeton, NJ