[bigshot]

  how on earth can anyone convience another in superiority of DSD - who does not have native DSD playback capability

 
Simple! By pointing to some double blind listening tests that show that people can hear a difference between DSD and redbook. Not hard at all.

 

[esldude]

  That is true. But how on earth can anyone convience another in superiority of DSD - who does not have native DSD playback capability and is thus forced to listen to DSD bounced down to PCM in any case ? He/she gets to hear whatever highest PCM resolution his/hers DAC is capable of - no true DSD/PCM DBT possible at all. 
 
I thought that DXD will be sounding a lot more close to DSD (whatever resolution/sampling rate ) - yet it does not. It still sounds somewhat cold and hard to me -  which is almost never a case with DSD. Funny enough, I found 192/32 bit floating bounced down from DSD128 the best sounding PCM to date.
 
Again, without DAC capable of native playing of all of the above, it is a guessing speculation act - at best. 

Again, without DAC capable of native playing of all of the above, it is a guessing speculation act - at best. 
 
Actually no it is not.   You think that because you adhere to the old high end audiophile method of subjective listening being the ultimate arbiter. 
 
In Sound science DBT results would convince as would pertinent measurements of audible differences.  Also since a good many measures of those formats are available there is a plethora of useful information so that it is not guessing speculation even though one may lack native DSD playback.  Also, you do realize many DSD dacs no longer adhere to one bit DSD, but use something like 4 bits to implement it?

 

[ab initio]

I don't think it is strictly necessary to directly compare DSD and PCM on native hardware. While that is the most direct method, one can use an extra step or two, using process of elimination to determine whether there is appreciable difference in the formats.
 
There's nothing wrong in doing a direct blind test between DSD and PCM on native hardwares, but there are many challenges that result in making sure that the different hardware isn't responsible for the test result. there are two variables being changed here, 1) the format and 2) the hardware. A correct testing procedure requires testing a single variable at a time. One would have to verify that the analog backends of the different DACs were of identical performance, and that the D-to-A processes in each DAC are accomplished with similar goodness.
 
By converting a native DSD recording to PCM, then reverting it back to DSD and comparing the native DSD track to the DSD-crippled-by-the-limitations-of-PCM track you can play back both tracks through the exact same hardware, thus eliminating that variable. The requirement here is that the conversion processes must be understood -- however, this is significantly easier to quantify because it's all math!
 
Conversely, one could repeat the test with native PCM vs a PCM-to-DSD-back-to-PCM configuration.
 
 
With regards to DSD<->PCM conversion. DSD is simply 1-bit, 2.8224 MHz PCM, right?
PCM upsampling and downsampling is very well understood.
 
So why can't DSD be converted to PCM via the exact same process? namely,

• take all of the samples (in this case, 1-bit) and convert them to high precision floating point numbers,
• upsample to a least common multiple sampling rate as the desired target PCM sample rate, 
• low pass filter at the nyquist frequency of the target PCM stream
• decimate to the sample rate of the PCM stream
• noise-shape dither and digitize to the target PCM bit depth (e.g., 16-bit)

 
The process of converting the PCM to DSD is exactly the same; namely,

• take all of the samples (in this case, say 16-bit) and convert them to high precision floating point numbers, 
• upsample to a least common multiple sampling rate as the desired target DSD sample rate, 
• low pass filter at the nyquist frequency of the target DSD stream
• decimateto the sample rate of the DSD stream
• noise-shape dither and digitize to the target DSD bit depth (i.e., 1-bit)

 
All of these can be done "mathematically perfectly". Well, at least to the precision of the computer arithmetic which far far far far exceeds any audio format precision. 
 
What fundamental concept am I missing here? 
 
Cheers

 

[ferday]

I

What fundamental concept am I missing here? 

Cheers

Sounds good to me...I'd be happy to make the files but I can't do native DSD so I can only do one portion of the comparison...

as i mentioned before, analogsurvivor had posted a link to "the same file" in a couple PCM formats and DSD. they are supposedly the same file, but i have no way of proving that. i did do ABX, but of course my system downsamples both the DSD and the ultra-high PCM to 24/96 so really the results are meaningless.

my STX Essence supports DSD but there is something in the chain which disallows actually playing it (apparently a DSP chip). Also, even in downsample the DSD file was easily audibly louder, and at the time i was in little mood to do more than ear-matching volumes since i knew the results were not going to be official

 

[analogsurviver]

  Again, without DAC capable of native playing of all of the above, it is a guessing speculation act - at best. 
 
Actually no it is not.   You think that because you adhere to the old high end audiophile method of subjective listening being the ultimate arbiter. 
 
In Sound science DBT results would convince as would pertinent measurements of audible differences.  Also since a good many measures of those formats are available there is a plethora of useful information so that it is not guessing speculation even though one may lack native DSD playback.  Also, you do realize many DSD dacs no longer adhere to one bit DSD, but use something like 4 bits to implement it?

First, it is my understanding that DBT is primarily listening test - not something to be measured and ticking the boxes that are officially recognized as being relevant to the perception of sound as officialy sanctioned by AES or any other relevant body. I stopped reading AES approximately at the time the algorithms later used for MP3s were all the rage - playing a direct to disc analog record against CD should be enough for anybody with ears -
throwing away yet more information as in MP3 and calling it science - no, thank you ! It is science - but one used for achieving much lower quality product while justyfing it with "most people can not hear or do not care about the difference".
 
DSD DACs can also use more than one bit to implement it - and these are NOT true DSD DACs. It is an area which more often than not receives "no comment" answer from the respective manufacturers.
 
One bright exception in this "no comment" DSD and PCM DAC land is iFi Audio - just check to what level of openess they went here on the threads at HF regarding their nano and micro iDSD DACs. Explaining why using a particular almost "vintage" DAC chip - while squeezing out of it MUCH more than it should have been capable of according to data sheet from the manufacturer. These DACs do both DSD and PCM natively and perhaps represent the most economic way of hearing both trough same hardware ; mere different analog output stage (as in dedicated PCM only and DSD only DACs ) can change the sound more than either DSD or PCM if both are implemented well.

 

[bigshot]

Please cite a double blind test where someone was able to discern a difference between PCM and DSD. I don't believe it.

 

[analogsurviver]

  Please cite a double blind test where someone was able to discern a difference between PCM and DSD. I don't believe it.

As I said, I will post the samples of recordings. Then you and everybody else can decide first hand. Let's say the samples will be posted by 15th August, possibly sooner. Should be enough participants to satisfy any statistical  requirements. If you are willing, I can let you, as the most prominent opposition, process all the incoming results via PMs. The condition for the participants to be able to participate will be a simple one: confirmed native DSD playback capability. We could add level matching to within less than 0.1 dB - but that is harder to do for the participants and yet harder for us to confirm the precision of their level matching .
 
What can you say regarding the above offer ?
 
You can go to any of the following links with free samples in various PCM, DXD and DSD resolutions and also listen/compare for yourself. However, as noted many times before, only native DSD playback will reveal the true difference. You will need another honorable person to help you with switching for DBT - foobar2000 (or anything else known to me for PC ) at the present can not play native DSD under ABX comparator. Things are developing in digital audio so fast at the moment that software will eventually have to catch up with the latest hardware. For example, mac enviroment at present does not allow DSD256 ( limited to DSD128 ) - PC/windows does. Foobar2000 even has provision for setting the DSD rate up to - DSD512 ( 8 times the resolution of SACD ).
 
http://www.2l.no/hires/
 
http://www.channelclassics.com/try-it-now
 
https://justlisten.nativedsd.com/albums/just-listen-1-compilation
 
http://audiogate.bluecoastrecords.com/
 
http://www.highdeftapetransfers.com/
 
etc etc
 
The problem at the moment is relatively very few users have native DSD playback capability. As any novelty, it is expensive at first - and as always, there will be fishing in murky waters. But that is changing - FAST. I do not want to come off as a fanboy of any manufacturer - but check for yourself what 189 $/EUR ( lucky Americans... ) can buy you today in DACs. Your last name no longer has to start with G- and end with -ates for you to be able to afford a very high quality piece of digital audio equipment.
 
Many have ( myself NOT included ) more money in a single cable connection...

 

[bigshot]

Just point me to a controlled test. I'm not interested in doing the test myself.

 

[analogsurviver]

If I were aware of such a test of DSD being perceived as clearly superiour to PCM - would I be offering to provide for another ?
 
Technology and its accesibility/use are reaching a point such a test could be made online. There is myriad of reasons why previous tests in this direction did not produce the result I am after. The leading one being the recording used that very seldom started its life as DSD in the first place.
 
I can maybe arrange for DBT among the musicians from forthcoming recording sessions - but based on previous experiences, there is little chance I could expect the required attention from them atop their main assignment - which is to rehearse and give best possible performance in the evening. I will try - but it is above my power to promise anything.

 

[castleofargh]

   
DSD DACs can also use more than one bit to implement it - and these are NOT true DSD DACs. It is an area which more often than not receives "no comment" answer from the respective manufacturers.

maybe I misread you but your sentence gave me the idea that you favored a one bit DAC.

 

[Roly1650]

Your non-sequitur diatribe about MP3 finishes with another of statements that listening to direct to disk analog vs CD should convince anyone.  As someone who has digitally recorded direct to disk analog and found nothing lacking, I am of the firmly supported position the audible performance envelope at least equals the analog disk.  Heck the old reel to reel exceeded LP fidelity, but somehow the LP was anointed king of the musical reference universe in some minds.

I did the same thing, a good few years ago now, with "I Got The Music In Me", supposedly one of Sheffield Labs better efforts, (Doug Sax). Recording from a then $8,000 turntable/arm/cartridge/step-up to a Sony bitstream cd recorder ($300). That's when I decided the digital format was every bit as good and the turntable got sold pdq.
I also agree 100% with your views on R2R, a Revox, Tandberg or Ferrograph, (remember them?) could run rings round any vinyl setup.
As an aside, I also had a Nakamichi real time duplicate cassette of one of the tracks from the Thelma Houston album which I also transferred to cd dubbing off a Nakamichi CR 7 and it was harder to set the levels off the cassette than it was the vinyl, which also indicates to me that the cassette was at the least every bit as dynamic. I keep meaning to run both the vinyl and cassette through Audio Diff Maker to see what the differences really are.

 

[bigshot]

I did the same thing, a good few years ago now, with "I Got The Music In Me", supposedly one of Sheffield Labs better efforts, (Doug Sax).

 
I did it with Lincoln Mayorga and Distinguished Colleagues Vol 2. Same result as you two. Indistinguishable.