This is something I explained in my review of the
Korg MR-1000, as well as in other technical SOS articles.
That Korg article is fundamentally flawed and false in several of its arguments and claims, and who ever wrote it doesn't understand the basics of linear PCM, let alone DSD. It is clearly a marketing puff piece, and sadly contains some serious misinformation (such as in the description of oversampling), and misdirection (such as the squarewave graphs), while also spouting unproven opinions as facts! Can you tell I'm less than impressed...?
However, the basic argument is this:
You can trade off reduced word length for increasing sample rate. Philips introduced this concept back in the early 1980s with their oversampling converters (4x oversampling enabled the use of 14 bit converters to deliver full 16 bit 44.1kHz audio quality.
As the technology improved, they introduced Bitstream which used 64x oversampling and a slightly bodged 1 bit output to provide the same quality. Their competitors offered similar solutions, such as MASH.
These techniques led on to the practical development of delta-sigma conversion techniques. Essentially, the analogue signal is sampled at a very high rate (~6MHz or more) but with very low word length (typically less than five bits, sometimes only one). A digital filter (decimation filter) process is then used to translate the information from high sample rate, low word length to lower sample rate, high word length for storage or further processing.
D-A conversion works the other way, with digital filtering (oversampling) to convert the low sample rate/high wordlength signal into a very high sample rate and low wordlength signal.
The advantages are far greater linearity (way beyond that achievable with 'conventional' converter designs as used in the last century), much reduced manufacturing costs, far better stability and accuracy, and a generally better sound because of all those factors (and many others).
Everything currently produced uses delta-sigma converters, because it is not practical or cost effective to do it in any other way... but Sony had the bright idea of simplifying things further.
They wondered why bother with the decimation filter in the A-D if you're only going to reverse the process again with the oversampling filter in the D-A? Why not just leave the data in the raw high sample rate, low wordlength form and save the hassle?
And that makes sense if you are only talking about a recording medium. So that's what they do in the DSD (and the SACD) format and on the face of it, it seems a good idea...
But as with all these things there are swings and roundabouts to consider. Firstly, and probably most importantly, many argue that the current 2.8224MHz DSD sample rate isn't high enough to support true 1 bit wordlength at the intended resolution -- and that's why the bigger Korg recorder offers a second higher DSD rate. Those in the know (and I'm talking about Sony's own egg heads here) argue that even that isn't high enough to do the job properly...
Secondly, huge amounts of noise shaping is required in these formats, and that results in a lot of out-of-audible-band energy that can cause problems in some cases. It also means that the oft cited claims of huge signal-noise ratio afforded by DSD only apply at very low frequencies. The noise performance gets progressively worse with increasing frequency. Linear PCM 24/96 has a considerably better signal-noise ratio at 40kHz than DSD... should such things be of concern to you ;)
It also makes signal processing (gain, EQ, dynamics, etc)very difficult indeed. Many DSD-based DAWs and other processing equipment actually converted back to PCM to do the work, and then back to DSD for the output -- thus completely negating all of the claimed advantages in one swoop! Later systems were developed to work more in the DSD native domain, but with less flexibility and even then the processing involved multi-bit stages -- it has too -- and even if only two or three bits further (simpler) decimationand oversampling filtering stages are still necessary.
The inimtable Dr John Watkinson wrote a very entertaining demolishing article about DSD in an edition of Resolution magazine a while back, too...
Personally, I'm quite happy sticking with 24/96 which I reckon is the best compromise in terms of audio quality, and the most practical in terms of processing, storage, and archiving.
But DSD does have its fans...
hugh
Edited to add: Don't take my word for it, read this abstract from two of digital audio's leading gurus:
Quote:
Stanley P. Lipshitz and John Vanderkooy
Audio Research Group, University of Waterloo
Waterloo, Ontario N2L 3G1, Canada
ABSTRACT
Single-stage, 1-bit sigma-delta converters are in principle imperfectible. We prove this fact. The reason, simply stated, is that, when properly dithered, they are in constant overload. Prevention of overload allows only partial dithering to be performed. The consequence is that distortion, limit cycles, instability, and noise modulation can never be totally avoided. We demonstrate these effects, and using coherent averaging techniques, are able to display the consequent profusion of nonlinear artefacts which are usually hidden in the noise floor. Recording, editing, storage, or conversion systems using single-stage, 1-bit sigma-delta modulators, are thus inimical to audio of the highest quality. In contrast, multi-bit sigma-delta converters, which output linear PCM code, are in principle infinitely perfectible. (Here, multi-bit refers to at least two bits in the converter.) They can be properly dithered so as to guarantee the absence of all distortion, limit cycles, and noise modulation. The audio industry is misguided if it adopts 1-bit sigma-delta conversion as the basis for any high-quality processing, archiving, or distribution format to replace multi-bit, linear PCM. |