"DSD,aka SACD,returns to multibit when dithered. Therefore only new recordings offer the advantages of true DSD"
---11-10-03: steve@sanibelsound.com
With all due respect, I feel a clarification is in order.
Dither is added to PCM digital audio to try and overcome the distortion that occurs at low nomerical values (quantization error). Now, it is true that when DSD is resampled into PCM, there is noise shaping involved. But let's not confuse the two. Noise shaping is employed to acheive a higher perceived resolution in the critical band (midrange) of PCM by shifting noise components into the ultra HF. Dither is randomized noise that is added to PCM in order to subjectively reduce quantization error that occurs at low numerical values (low volume portions of the audio).
Now, this issue about "pure" DSD recordings. Due to the mathematical properties of the 1-bit system, and the fact that there is no "word" at all in DSD, all DSD recordings edited, mixed and or mastered in the digital domain on DSD Digital audio workstations have to undergo a transformation into some sort of a multi-bit audio stream whenever any sort of digital signal processing is involved. Processing in this case can mean something as simple as pulling a virtual fader to change gain, the use of dynamics processing, or the implementation of EQ. The PCM involved is extremely high resolution [64fs in most if not all cases (64 x 44.1kHz or 2.8MHz!)] You can also think of it as almost 30 times more samples than the 96kHz DVD-A standard. However, here is where Noise Shaping comes in. Here is also where the controversy of SACD's dynamic range in the high frequency extremes comes into play.
So what is pure DSD? Unless you've been at a DSD recording session/demo that played back unaltered DSD, or you've heard an SACD that didn't undergo any processing or that involved active mixing in a DSD workstation, you have not heard pure DSD.
Does this matter? Not really. It's all just numbers on a page at this point. Both DSD and high resolution PCM have their technical issues. Both offer a huge improvement in resolution over the 16 bit 44.1kHz standard. It's also practically impossible to compare the sound qualities of both of these modulation schemes. The DAC's employed in PCM are dramatically different (electronically) than the ones employed in DSD. Even if you have a DAC that converts both PCM and DSD (such as Meitner or Genex), it's hard to make real comparisons. You end up hearing the implementation of the different D to A circuitry. It's still apples and oranges.
So, who wins? You pick. SACD seems to have a jump start on sales of both hardware and software. DVD-A has some cool video features.
Me? I like DSD and SACD. It's just such a mathematically cool way of looking at digital audio. It's almost zen... but I'm also a huge nerd...