There's a key difference between the DSD 1-bit format (at either 2.8 MHz or 5.6 MHz) and ALL the other digital formats (though with DXD the difference blurs) that doesn't seem to be mentioned on this forum. (If I'm wrong about this I'd welcome a correction, partial or total, by the way.)
All the digital formats, except for 1-bit (which was originally commercialized by dbx incidentally, as an add-on unit for video recorders: 700 kHz/1-bit), are based on the Nyquist theorem, developed in the 1920s, for telephone use. The Nyquist theorem in turn was based on the theorem on vibratory motion developed by the French physicist and mathematician Jean Fourier (1768-1830). The Fourier theorem states that all combinations of oscillating sounds — meaning sounds of REGULAR PERIOD — can be analyzed as combinations of sine waves (the famous "Fourier Transform" , on which all PCM, non-one-bit digital recording and reproduction is based.) PCM, non-one-bit, recording — which uses a chip to, in effect, disassemble complex musical sounds into combinations of sine waves; and then uses a second chip (in playback) to "reassemble" them into sine waves works perfectly on musical sounds — IF THEY ARE RECORDED IN ANECHOIC CHAMBERS, AND IF THE TOPS OF STRINGED INSTRUMENTS (AS WELL AS MICROPHONE STANDS, ETC.) ARE PADDED WITH ABSORBENT MATERIALS SO THAT THEIR ARE NO REFLECTIONS FROM ANY SURFACE WHATSOEVER. The sine-wave based system is also one that works well for telephone use, where tone, timbre, "air", etc. are non-significant. Incidentally, the sine wave is the most common (and, I'd say, elegant) form of motion in nature — it's also, of course, called "harmonic motion." But the problem with using sine waves as the electronic "medium" to record music in real spaces, rather than in anechoic chambers, arises from two facts: 1) musical instruments have, since the mid-1700s, been built to play and/or played to an equal-tempered scale (piano, organ, etc and fretted instruments can only play in equal temperament; other strings and winds and voice are adapted in performance to play with them. The strings on, for example, a piano know nothing of equal temperament, so their natural overtones are slightly out of tune with the other strings on the instrument, and so create beat tones, which in turn have their own natural overtone series, all the overtones beating against each other. Helmholtz, by the way, hated organs, because he couldn't stand this kind of unavoidable beating in tempered instruments. When you COMBINE this unavoidable BEATING of well-tempered instruments whose individual strings and columns of air cannot possibly be tempered WITH the PHASE DELAYS created by sounds reflected from the physical bodies of instruments, chairs, walls, moldings, etc. in any normal music environment, you get sounds that are APERIODIC, or, more exactly, aperiodic within any conceivably recordable and reproducible frequency spectrum. Thus, they cannot be analyzed, stored, and reproduced as sine waves. These are very low-amplitude, high-frequency sounds. When CDs and other Nyquist-based digital formats are said to lack "air", what is really being referred to are those very low-amplitude, high-frequency sounds that reveal the _presence_ of air in the space where the music was recorded. 1-bit, DSD, recording doesn't try to analyze sound as sine waves, as periodic sounds, and so this intractable problem is avoided. That's why there's no D/A converter at the end of the playback chain (which I call, in the case of CDs, a "sine-wave pump"), but only a low-pass filter. Around the time CDs first became available (c. 1980) the AES Journal ran an article with photographs of all the sounds "between the samples" — fast spikes, for example — that CDs ignored. These, and others, are the sounds I'm referring to. Recorded music engineers and electronics designers, by the way, have developed all kinds of excellent tricks to disguise the absence of non-periodic musical sounds: they are for the most part variations of the Aphex sonic enhancement approach of the mid-'70s: adding extra harmonics, at lower frequencies often, to add INTEREST that the ear misses: the brain wants to know where sound is coming from, and it wants to work at that. With all respect for those who enjoy multi-channel music, I wonder how many of them would still appreciate multi-channel if they hadn't been deprived of full musical enjoyment for a sufficient time by Nyquist-based CDs, not to mention MP3s.
|