mike_mccue
while you can probably find lots of CD players that have comparatively crappy sounding output sections.

But that's assuming you're using an analog out on the CD player.
 
And what if the SACD (DSD) player conveniently converts the DSD to PCM when you digitally connect it to something that doesn't do DSD. 
 
 
 
Really there's two separate points here:
 
1. DSD provides "higher resolution" (in terms of numbers) than CD quality 16bit/44.1 kHz. There are many arguments about whether this is audible under any sort of typical circumstances and sometimes the people with the least knowledge have the strongest opinions, and many people would rather argue than reason.
 
2. Assuming "higher resolution" is desirable to someone, there is more than one way of achieving this. In the consumer world, SACD uses DSD while DVD-A and Blu-Ray audio discs use higher resolution PCM (often using lossless compression).

.

One issue is DSD is not exactly ASIO compatible.  So forget using a USB device or a boutique ADC device like Merging Horus or DAD for native DSD recording into Sonar, at least not yet.  You can use devices like the Tascam to record from the high end converters and import that file later on.  Though I have owned the original DV-RA1000, and thought the conversion quality was excellent by itself.  

2. Assuming "higher resolution" is desirable to someone, there is more than one way of achieving this. In the consumer world, SACD uses DSD while DVD-A and Blu-Ray audio discs use higher resolution PCM (often using lossless compression).

 
Even DVD-A uses specialized players as it is not a typical format, a real PITA.  Blu-Ray audio is my favorite of the new formats, because basically is just a typical Blu-Ray with AES standardized menu option that can be operated with or without video.  Allowing for multi-channel, high resolution PCM recordings anyone can produce, distribute, and listen to with inexpensive consumer equipment.  Of course, I think the future is still the internet.  As longs as we can get away from compressed formats.  

.

There are more aspects to DSD that are all important.   It's easy to pick on the first one and say "but that doesn't matter".
 
1. Higher sample rate - in of itself is not what makes the audio sound better.
2. Noise shaped.  It's one-bit audio, that'll sound bad if you do nothing to it, but part of the DSD process does noise shaping so you get a super low (below -120dB) noise floor in the traditional 20K audible range.
3. Signals can go over PCM 0dBFS - that makes it less important to do brick wall limiting during the mastering, and you end up with recordings with more dynamic range, or less distortion.
4. Higher processing requirements mean that the equipment is made to a higher standard, and will use better parts throughout.
and then there's PCM's dirty little secret :
All audio class PCM A/D converter chips are actually Delta-Sigma samplers with a built in a D-S to PCM converter.
The cost of the chip directly affects how complicated they make the D-S to PCM converter, and hence its quality.
By ignoring the part of the chip that is made to a price point, and grabbing the data directly at the Delta Sigma sampling point, we get a DSD stream.  Then we can use software to do the DSD to PCM conversion at whatever quality we want, even throwing the full 64 bit floating point resources of an Intel hex core i7 at it  Neither a  $1 A/D chip nor a $5 A/D chip can match this.
A DSD D/A chip similarly bypasses the internal up-sampling that goes on in a PCM chip, and drives the final output stage directly. 
 
It's a case of less is more...
 
p.s. editing at "DXD" 384kHz/24bit means you can keep the shaped noise in the signal, there's nothing audible up there - but the pre-existing shaped noise makes the conversion back to DSD sound better.
 
Tom Duffy (TASCAM)

There you have it from the source :) Thanks Tom for the expert perspective on DSD.

I am very interested where Sonar takes DSD. 
Tascam have a DSD stereo recorder. Korg were another that produced one.
Some have several sinc'd via s/pdif to form a multitrack machine. 
A Tascam DSD converter and Sonar DSD multitrack replay would be cool one day.
 

A little more about DSD is it is a more linear method of recording digital and works by sampling the difference of 1 bit to the next 1 bit sample at up to 5.6mhz times per cycle. 
PCM samples the total bit length at every sampling frequency cycle.  It can be graphed as a step. 44.1 khz at 16 bits for eg. 
 
 

Tom's post covered the most potentially relevant part "Then we can use software to do the DSD to PCM conversion at whatever quality we want, even throwing the full 64 bit floating point resources of an Intel hex core i7 at it  Neither a  $1 A/D chip nor a $5 A/D chip can match this."
 
There is no inherent advantage to PDM like DSD uses over PCM. They are just two different ways of describing a wave in discrete time. However there is an advantage in using them for actual input and output. It is hard to the point of near impossible to build an actual 24-bit DAC, as in something that would produce over 16 million discrete voltage levels. So it isn't done. Instead, you use a 1, 2, or 4-bit DAC that operates at a much higher frequency. The math for doing that is very well understood and established and it works great. However there is the decimation step where you convert from one to the other, and that isn't necessarily as high quality as you might want. Limited budget for transistors and all that jazz, plus it has to happen in realtime.
 
With DSD you eliminate all that and just store a 1-bit high frequency signal. Solid theory at any rate.
 
I don't know that in practice it makes much difference these days. I would question if good converters would have an audible difference. However it can't hurt to have as an option for sure.
 
Do be careful with regards to information though as there is more than a little snake oil and zealotry floating around about it. You'll hear people start to talk about the "purity" of DSD and so on, failing to understand that it is all math in the end :).