pdh (12/15/82)
For those of you interested in the digital audio debate, I post the following scientific (not overly technical), but nonetheless uncorroborated information: In the process of digital recording, as the analog signal is sampled at roughly 50KHz, saved, and then reconstructed through digital-analog converters, a problem arises when working with signals near the 20KHz hearing limit of the human ear. Since at 20KHz the A/D side of the system has time to make only about 2 samples of the input signal, which is ASSUMED to be sinusoidal, the reconstructed signal is generated as a sinusoidal waveform. This is not necessarily the original signal, and, in the case of the human voice for example, will be offset from the logical zero-crossing by some amount, since the human voice generates sine waves that are mostly above the zero crossing (i.e. the vocal chords spend more time in one direction than they doo in the other). Whether the zero-crossing offset problem is audible is subject to question. However, regenerating an unusual (i.e. non-sinusoidal) waveform as a sine wave is definitely audible -- the subharmonics are completely different. The counter-argument to the above problem is that the speakers that music is played through act as low-pass filters, and therefore a square wave (for example) will get reshaped into a sine wave; unfortunately for this argument, though, is the fact that there is a documented audible difference between a sine wave of frequency L, and a non-sine wave of the same frequency... The other problem in digital audio is much more subtle, yet may well be the cause of the much-fabled ear fatigue that can occur with digital audio recordings... One critical element of any stereo recording is the imaging; in a good recording, you should be able to place the instruments in their approximate locations. In digital audio, the imaging can be thoroughly butchered by the fact that the sample-and-hold circuitry of the two channels are rarely in perfect synchronization. This means that there can be a significant phase difference -- upwards of 10 degrees or so -- at high frequencies, where the critical imaging information is. Worse yet, since the locking speed of the sample-and-hold circuitry is signal-sensetive, it is possible for the phase misalignment to change with signal amplitude, causing the audible image to "waver" within about 10 degrees in the listening field (which itself is usually only a fraction of a circle). So, simply put, with the possibility of such significant image fluctuation, it is not hard to imaging tremendous ear fatigue, as your brain tries to "stabilize" the audible image. While it is likely that these extremes of phase misalignment will not occur, I can say that I have heard this phenomenon occurring within an audibly significant degree range. It seems to me that the solution to both problems, if you want to keep the inherent advantages of digital, is to wait for the technology that will allow about a 200KHz sampling rate, possibly more. Peter Henry hplabs!pdh PDH@SAIL (415) 857-5921
burris (12/16/82)
#R:hplabs:-104500:ihlpb:4000025: 0:131 ihlpb!burris Dec 16 13:56:00 1982 Well, here we go again! This conversation seems to repeat itself over and over again. Dave Burris ihlpb!burris BTL - Naperville
tony (12/18/82)
#R:hplabs:-104500:pur-ee:12000010:000:114 pur-ee!tony Dec 17 10:47:00 1982 I give up. When people start telling me that I can hear DC, it`s time to stop reading net.audio Tony Andrews