DXD versus DSD

[awsmone]

intersting article about dsd quality degradation with DXD passage

http://www.stereophile.com/content/promates-worlds-first-dxd-download-store

 

[amirm]

I read the original article on their web site earlier today! As marketing material go, it is not half bad .

I had fun reading the comments including the correction from our own Bruce Brown.

This last comment though was funny:

There is no benefit for these high sampling rates. We cannot hear or feel them.
There is an exception if you are an audiophile german shepherd.

I had not read that line before .

 

[Orb]

Is there any reason why they would not sell the music in a compressed archived format such as RAR?
Thought that would make sense, anyone messed around to see what saving one gets with 24/192 album or DXD (if downloaded from them)?
Late here and a fair amount to do so unfortunately will not get a chance to mess around with it myself.
Context here is not about streaming but those who buy/download to play tracks and albums.

Cheers
Orb

 

[Bruce B]

Peter has been a big proponent of DXD when it was first extablished as the editing mode for DSD. I will have to say he is incorrect when he starts talking about UHF frequency building up and starts exceeding Scarlet Book spec. This was true 10yr. ago. That's why we now record at 2x and 4x so we won't have that build-up and can do multi-track mixing or multiple conversions.

 

[awsmone]

thats interesting Bruce

this would appear to breakdown their whole raison d'être

any graphics to back that up per chance?

 

[Bruce B]

thats interesting Bruce

this would appear to breakdown their whole raison d'être

any graphics to back that up per chance?

You really don't need graphs. The UHF noise in DSD64fs starts around 24k and rises from there.



DSD128fs pushes that noise out one octave where it starts at about 48k and so on for DSD256fs.




Now the Scarlet Book spec calls for only a certain amount of UHF noise can be had from 20k-50k and from 50k - 100k. You can see how we measure and make sure we don't go over the limit here on the Sonoma screenshot. Here the UHF is exceeding the threshold that is allowed for 50k - 100k.

 

[Orb]

Peter has been a big proponent of DXD when it was first extablished as the editing mode for DSD. I will have to say he is incorrect when he starts talking about UHF frequency building up and starts exceeding Scarlet Book spec. This was true 10yr. ago. That's why we now record at 2x and 4x so we won't have that build-up and can do multi-track mixing or multiple conversions.

Maybe I am misunderstanding but that link says to me that one should use DXD instead of DSD to avoid the noise rather transcode from DSD to DXD; maybe this position has changed from the past.
His current example reminds me of the issue of using dither and reiterating that multiple times at mixing-editing; DSD is great until one has to transcode it to PCM for some nifty edits-mixing (sure someone will mention dsd-wide resolves all issues ).

I agree there seems quite a few DSD recordings that do not need to be transcoded to PCM, but this ties the listener to a specific genre of classical and still limited releases, anyone know how many modern jazz recordings are straight DSD to mastering as I assume that makes sense to try as well?

Bruce when you record at 2x and 4x you still have the shaped noise, the theoretical-technical issue is the sum of that high frequency noise as voltage output from the DAC and its potential influence on wideband preamps/amps; this is where the noise is noticeable.

Cheers
Orb

 

[FrantzM]

Is there any reason why they would not sell the music in a compressed archived format such as RAR?
Thought that would make sense, anyone messed around to see what saving one gets with 24/192 album or DXD (if downloaded from them)?
Late here and a fair amount to do so unfortunately will not get a chance to mess around with it myself.
Context here is not about streaming but those who buy/download to play tracks and albums.

Cheers
Orb

Not sure .rar would be a good fit for music. Same goes with .zip. Thus the flac, .ape. vw of this world

 

[Orb]

Not sure .rar would be a good fit for music. Same goes with .zip. Thus the flac, .ape. vw of this world

Flac is more a realtime-streaming solution where as rar is a heavy offline compression one, but you could be right (although thinking more of winrar in this instance).
Of course there are other heavy offline multi file archive compression algorithms (like rar) that are more open standard (or used more by open standard community although rar is pretty much heavily used).
No-one done a comparison?

Edit:
Also mentioned rar for greater flexibility in terms of multi file archive and also usable for all music files such as DSD; how much it compresses is what would be interesting.
That said surely any commercial audio digital download sale should use a solution with some kind of compression to make them much smaller, although not all do (I assume).
Just to add, does FLAC support encoding of DXD? ---> Should do as within sampling rate limits *shrug*, so not sure why some sites offering DXD are not also offering it as FLAC.

Cheers
Orb

 

[Kees de Visser]

FLAC gives me an average compression ratio of about 0.65 with DXD files.
Besides the data reduction there's the advantage of checksum verification.

 

[Orb]

FLAC gives me an average compression ratio of about 0.65 with DXD files.
Besides the data reduction there's the advantage of checksum verification.

You get that with nearly all archive compression solutions though.
Their benefit is keeping it as WAV, although some may see that as detrimental due to hard disk storage requirements (may be a consideration if large library on SSD drives)
Thanks for headsup on ratio using FLAC.

Cheers
Orb

 

[Orb]

Hmm well seems not easy to compare when going outside of FLAC as results may be more varied, as an example I swear some of the RAR compression software (such as WinRAR) had good compression for WAV but quick n dirty testing for me had winzip giving much better results (still only 33% compression); caveat being I only did quick and dirty testing without fine tuning.
So guess comes back to APE as Frantz mentioned - or alternative such as OptimFrog.
Software thinking from top of mind is Monkey's Audio for ape that compresses slightly better than FLAC (could be wrong here).

Cheers
Orb

 

[Orb]

Just to add, was interesting downloading from 2L, really need to use some kind of download manager otherwise will take 10x longer (from a quick test I just did).

Cheers
Orb

 

[microstrip]

As far as I remember, one the most promising 2014 DACs having DXD and DSD was the new Nagra HD DAC.http://www.nagraaudio.com/hd-dac/ Does any of our members have experience with it?

 

[tailspn]

Maybe I am misunderstanding but that link says to me that one should use DXD instead of DSD to avoid the noise rather transcode from DSD to DXD; maybe this position has changed from the past.

Yes, I read it the same way. What the author doesn't understand, and Morten doesn't tell, is that there is no way to record in DXD. There's always a delta-sigma modulator (think DSD or DSD-Wide) front-ending all DXD capable A/D converters. All DXD content is a conversion from either 1-bit, or multi-bit Pulse Density Modulation bit streams, either within the A/D converter in realtime, or after the fact in production editing/mastering.

It eludes me how one can sell the idea with a straight face that decimation filtering, and best curve fit math conversion can improve the quality of a analog sourced DSD/DSD-wide recording.

On the other hand, as the Grimm DSD Myth white paper:

http://www.grimmaudio.com/info/news/all/all/4799

details, if DSD/DSD-Wide content requires post processing sweetening, which is most often accomplished in DXD, then the resulting product should also be released in DXD, rather than only being converted back to DSD.

 

[Joe Whip]

Seems to me that DXD is PCM. End of discussion. It will take a lot to convince me that these higher and higher sampling rates make any difference at playback. Considering their cost over lower sampling rate files, I would expect much greater fidelity. As my DAC doesn't do DXD, I haven't experienced it but really doubt we will be abe to descern a difference. What is next, 64/784? This is starting to get absurd. I am sure all of this is well intentioned but given the cost, I am not so sure.

 

[tailspn]

Seems to me that DXD is PCM. End of discussion. It will take a lot to convince me that these higher and higher sampling rates make any difference at playback.

Well, you're right. DXD was not meant to be a 24/352KHz PCM release format, It's a professional post processing format for DSD recorded content. And yes, 704KHz and 1.4MHz PCM are just month's away, again as a post processing tool for 128fs and 256fs DSD recorded content. The 8:1 difference from DSD 2.82MHz bit rate to DXD 352KHz PCM sampling rate becomes 16:1 for 128fs DSD, and 32:1 for 256fs DSD today; a much less than ideal constraint on the decimation filters involved. Hence the new DXD rates to retain the 8:1 fold down ratio.

But audiophiles want the best obtainable, therefore industrious labels will attempt to fill it.