Topic: DSD playback

Hi

First post so please be gentle.

I am currently using Roon (NUC running Rock) to upsample all music to DSD256 and then streaming via an Auralic Aries G1 to my ADI-2 DAC. Music is either on a hard drive in the NUC or streamed from Tidal. I have a passive pre amp and active ATC speakers so not using the DAC's volume control.

I just wanted to check if this is the best thing to be doing? DSD256 sounds good and more 3 dimensional than other formats but I am not sure if I am losing anything (ability to use filters in the DAC?) by doing this.

Any thoughts gratefully received

Thanks

Steve

2 (edited by Curt962 2019-02-03 22:00:51)

Re: DSD playback

Steve!

Welcome!   There's a lot of Fine Folks here with Great Knowledge of the Technology.  You can only Win.  Enjoy your new DAC

All the Best!

Psst!  You're gonna catch some Flak though regarding Upsampling everything to DSD. 

Upsampling doesn't "make it so"   You will simply have a much larger version of the Original File, flawed as it may be.

Just Sayin!

Curt

Re: DSD playback

Hmmmmm I did wonder about that but then why upsample at all?

I know the PS Audio DSD DACs upsample everything to DSD so there must be some benefit.

4 (edited by ramses 2019-02-04 06:54:39)

Re: DSD playback

stidge wrote:

Hmmmmm I did wonder about that but then why upsample at all?
I know the PS Audio DSD DACs upsample everything to DSD so there must be some benefit.

First of all, please no emotional discussion of this topic. Here only my $0.02 on this topic:

If you search this forum for "upsampling" and "MC" then you find a few threads with "upsampling" in it.

Reading the forum regulary I understood it this way, that hardware and software upsampling are different in capabilities.

There was one statement from MC, which I don't find anymore (maybe because in this response the word "upsampling" did not occurr), where he said, that (HW?) upsampling will deliver the same quality as the source material but not more.

So HW upsampling in an ADI-2 Pro is there to make it possible to integrate devices with different (fix) sample rate into a setup with another sample rate, because otherwise it's not possible to operate those devices together.
But you won't get any benefits in terms of sound / quality.

In regards whether SW upsampling delivers better audio quality I remember no specific statement on this forum.

I am personally very skeptical to this because it also makes no sense to me that a source material of a certain sample rate / quality can be made better by SW upsampling. The quality or dynamic that is not there from the beginning can not be added later. Once something has been stripped down from maybe 88.2/24 to i.e. 44.1/16 can not be made better by this.
In my opinion it's only a clone of the 44.1/16 quality in higher resolution but with no added value, even worse, file sizes become much bigger for most likely no reason.

I think such a dedicated SW makes only sense sound wise for the other use case of "downsampling", from i.e. 96/24 to 44.1/16. Some people say that a dedicated SW solution for downsampling would have the better algorithms compared to a DAW product for mixing and mastering. Whether this is really the case I can't judge.

If you "think" it sounds better, then a general warning / advice.

1. Pls. be aware of psychoacoustic effects, that your ears are not the same every day or when having longer listening sessions. And our brain can not really remember sound. So you need to perform blind tests to avoid psychoacoustic effects and A/B tests within seconds.

2. only because a vendor offers a certain solution it does not necessarily mean that it brings real benefits, similar to audiophile USB cables, etc (we had this discussion already last recently in the forum).

3. it's anyway questionable whether DSD brings audible advantages (and I made myself the experience, how hard it is to hear differences between DACs or different digital filter modes of one DAC, even if you can A/B.

According to the german Wikipedia entry about DSD (https://de.wikipedia.org/wiki/Direct_Stream_Digital)

"The amount of data to be stored in the DSD format is generally larger than in the commonly used PCM format, but this does not necessarily lead to noticeable sound improvements. So far, there is no practical proof that the sound improvement of DSD over PCM claimed by developers and users actually exists. In a study conducted by the Hochschule für Musik Detmold, participants could not hear any statistically relevant differences between the data formats in blind tests. The authors of the study draw the conclusion that "even with the highest quality equipment under optimal listening conditions and a wide variety of hearing focus settings and sound quality levels, the data formats of the different data formats are not statistically relevant". The test persons' hearing experiences are usually not audible in significant differences between DSD and High Resolution PCM (24 bit/176.4 kHz), therefore it could be argued that none of the tested systems stands out due to sound characteristics" and refer to "the high degree of frustration that many test persons, most of whom were accustomed to professional and critical-analytical hearing, felt during the performance of the tests and that they attributed to sound differences not nearly recognizable to them" (http://www.eti.hfm-detmold.de/lehraktiv … t-176-4khz).

It can be assumed that the methods used in modern audio converters, such as oversampling and phase modulation (dithering), are sufficiently good for both recording and playback to prevent the potentially adverse effects of early decimation of the data rate to a low sampling frequency from becoming audible. In particular, the filter technology in the chips has made great progress in recent years. As a result, the amplitude and phase errors generated by the reconstruction filters are predominantly in the high-frequency range, which is only slightly occupied by typical audio signals.

Today, sampling rates of 192 kHz are used at 24-bit resolutions, which are not only much more accurate than the original format, but also allow less steep filters with fewer errors to be used. It can be mathematically demonstrated that a relatively softly flowing reconstruction filter, which is driven at 192 kHz sampling rate and therefore does not have to block fully until 96 kHz, can easily transmit linearly up to well over 20 kHz, and thus only theoretical improvements can be made if it were operated at even higher rates. Ultimately, there are no significant losses due to the data pre-filtering integrated in the chips.

If a DSD data stream is present, it is much easier and more technically possible to decimate to any PCM sampling frequency than is possible from existing PCM data of other sampling frequencies, since no resampling is then required. This is why such DSD data is sometimes used in samplers for music generation. Sample rate converters also use the DSD format as an intermediate level by first going up from a low sampling frequency, then filtering the DSD data similar to analog data and finally resampling or decimating it to the target sampling rate.

From a technical point of view, a great advantage of DSD is its simplicity and elegance in processing different sampling rates, as it does not require the auxiliary PCM filtering techniques mentioned above.

Translated with www.DeepL.com/Translator"

BR
Ramses
X10SRi-F, Win7 Prof, Cubase 9.5, UFX+, Octamic XTC, ADI-2 Pro FS BE/DAC, RayDAT, ARC USB

Re: DSD playback

Very Well Documented Post Ramses.  Viele Dank!

Re: DSD playback

upsample pcm44 to dsd make sense cuz you remove weakness of pcm filters near Nyquist freq
its like nos mode without aliasing

Re: DSD playback

dir wrote:

upsample pcm44 to dsd make sense cuz you remove weakness of pcm filters near Nyquist freq
its like nos mode without aliasing

But AFAIKthe same 44k playback filters must be used when upsampling, also in software upsampling, so there (upsampling) you end up with the same choices and compromises as with playback.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
HDSP9652+ADI-8AE  2X HDSP9632
Cubase pro10

Re: DSD playback

not excatly. for example:
Jriver (player), Tascam Hi-Res Editor (software) and TEAC UD-505 (DSP) uses three absolutely different algorithms for PCM->DSD conversion.
Jriver gives all the best on my taste and measurements also. Impulse looks like NOS without any ringing.