DACs : native DSD vs. native PCM

[amirm]

This is precisely what's confusing me. If the original source files are usually NOT DSD or else are converted DSD just for the purpose of being played through a DSD DAC, what's the point of getting DACs with native DSD processing chips?

There is a belief, expressed in countless forum posts and blogs that conversion of PCM to high-rate DSD makes the music sound better. The idea being that "DSD is more like analog" and as I mentioned, the bigger the number, the better and DSD rates are much higher than PCM.

There is no controlled listening test to back this impression. But it lives on in the world of subjectivity where anything goes.

The first step in analyzing this belief is to measure the difference. As far as I know this has not been done. So I aim to remedy that by measuring the actual performance of the DAC with and without this conversion. Once we have that data we can speak more authoritatively than we do now which is just to express our belief just like them.

If you ask me what I think, I say that volume can change in the process of such conversions and if so, that can easily lead to wrong conclusions regarding audibility. So that needs to be measured and confirmed to be the same. Next we need to determine if something good has happened as a result of conversion using measurements. If there are negative consequences then the case others are making becomes much harder. If there is an improvement then we need to analyze it to determine audibility.

Of course the ultimate test would be a blind listening test and I know of one formal, published test of this. I have done a digest of it on ASR forum. Click on this search and go to the first link to read it: https://www.google.com/search?rlz=1.......0...1.1.64.psy-ab..0.0.0....0.IU8BukXjygE

 

[amirm]

It seems like perhaps improved formats of PCM are more the future rather than DSD, so it's not like I'll be "future-proofing" by getting a native DSD DAC. From Wikipedia:
https://en.wikipedia.org/wiki/Direct_Stream_Digital
"DSD met with relatively little success in the consumer market, even though the SACD was actually more successful than its direct competitor, the PCM-based DVD-Audio. Direct manipulation of recorded DSD data is difficult due to limited availability of appropriate software. The advent of new high-resolution PCM standards, such as DXD, further restricted its market niche. DSD, however, is still used as an archival format for studio applications, and it's seen as a possible replacement for analog tapes." Though I'm not sure if accessing this studio archival material is easy for any given album I want to get.

Well, if you fast forward to a day where CD is in deep decline, then some of your music consumption will come from digital downloads. There, I like to download the version of the file in which the original recording was made. If that is DSD, I rather get that and play it in its native format than having them convert it, or me converting it. There are a lot of audiophile labels that believe DSD is superior to PCM (correctly or incorrectly) and capture their recordings using DSD. For this reason I recommend that you get DACs that accept DSD format.

I have fair number of DSD originated high-res content that I have purchased already.

The Wiki article is behind times based on what you quoted in that we are no longer talking about SACD but digital downloads. And there, DSD is quite alive and well. One good example and source of very well recorded music is Challenge Classics: https://challengerecords.nativedsd.com/

 

[TidalWave]

There is a belief, expressed in countless forum posts and blogs that conversion of PCM to high-rate DSD makes the music sound better. The idea being that "DSD is more like analog" and as I mentioned, the bigger the number, the better and DSD rates are much higher than PCM.

There is no controlled listening test to back this impression. But it lives on in the world of subjectivity where anything goes.

The first step in analyzing this belief is to measure the difference. As far as I know this has not been done. So I aim to remedy that by measuring the actual performance of the DAC with and without this conversion. Once we have that data we can speak more authoritatively than we do now which is just to express our belief just like them.

If you ask me what I think, I say that volume can change in the process of such conversions and if so, that can easily lead to wrong conclusions regarding audibility. So that needs to be measured and confirmed to be the same. Next we need to determine if something good has happened as a result of conversion using measurements. If there are negative consequences then the case others are making becomes much harder. If there is an improvement then we need to analyze it to determine audibility.

Of course the ultimate test would be a blind listening test and I know of one formal, published test of this. I have done a digest of it on ASR forum. Click on this search and go to the first link to read it: https://www.google.com/search?rlz=1C1CHFX_enUS591US591&ei=Np2ZWsKkNo-ejwOAtafICw&q=site:audiosciencereview.com+amirm+DVD-A+SACD&oq=site:audiosciencereview.com+amirm+DVD-A+SACD&gs_l=psy-ab.3...11305.11872.0.12236.5.5.0.0.0.0.56.253.5.5.0....0...1.1.64.psy-ab..0.0.0....0.IU8BukXjygE

Well, if you fast forward to a day where CD is in deep decline, then some of your music consumption will come from digital downloads. There, I like to download the version of the file in which the original recording was made. If that is DSD, I rather get that and play it in its native format than having them convert it, or me converting it. There are a lot of audiophile labels that believe DSD is superior to PCM (correctly or incorrectly) and capture their recordings using DSD. For this reason I recommend that you get DACs that accept DSD format.

I have fair number of DSD originated high-res content that I have purchased already.

The Wiki article is behind times based on what you quoted in that we are no longer talking about SACD but digital downloads. And there, DSD is quite alive and well. One good example and source of very well recorded music is Challenge Classics: https://challengerecords.nativedsd.com/

Thank you for your excellent replies! I fully agree with you on the first one, and the second gives great practical advice. I will read the article you've linked to.

Are there processors in mobile DACs that do "native" PCM decoding and also offer DSD via conversion? To be contrary to those that do native DSD.

 

[bigshot]

I don't know why you would want DSD files on a mobile player anyway. It would just take up a ton of space for no audible improvement in sound quality. If you have a DSD just transcode it to something more appropriate. There shouldn't be any audible difference.

The main advantage of DSD on SACDs is multichannel audio support. 2 channel DSD is pretty much pointless for home playback of recorded music.

 

[TidalWave]

I don't know why you would want DSD files on a mobile player anyway. It would just take up a ton of space for no audible improvement in sound quality. If you have a DSD just transcode it to something more appropriate. There shouldn't be any audible difference.

The main advantage of DSD on SACDs is multichannel audio support. 2 channel DSD is pretty much pointless for home playback of recorded music.

There are very compact mobile DAC/amps that have "native" DSD decoding, which is a reason I started this thread. For example:
http://www.tpdz.net/en/products/nx4dsd/index.htm
http://www.fiio.net/en/story/593

 

[amirm]

Are there processors in mobile DACs that do "native" PCM decoding and also offer DSD via conversion? To be contrary to those that do native DSD.

I don't think that there are any even in desktop versions. Ladder DACs tend to be PCM only. If you want to play DSD you connect them to a computer and do the conversion there.

 

[TidalWave]

I don't think that there are any even in desktop versions. Ladder DACs tend to be PCM only. If you want to play DSD you connect them to a computer and do the conversion there.

Ok, so by this logic, all mobile DACs do native DSD if they are specced to process DSD? (I'm probably missing some other native formats, so that's probably incorrect reasoning). So those that are not specced to decode DSD don't have native DSD decoding chips but rather most likely PCM? I wonder if I want the latter for the near future if I won't be using DSD anyway. Or if I should ignore DSD/non-DSD decoding specs and just go for other parameters of a DAC/amp, such as noise, amp power, reviews of "soundstage" etc.

 

[PurpleAngel]

That’s the path I’ve been taking: FLAC + 320kbps highest quality mp3 rips for each CD. That’s only for CDs: everything else I’ve accumulated over the years is variable quality mp3, which is what I’m stuck with to some extent. So I don’t expect miracles out of a DAC, but I’d like to understand if there are better DACs to look into. Some that are marketed for DSD playback have set me on a research path about DSD.

With the more poor quality digital audio files, a high quality DAC will just bring out more imperfections in the audio
Nothing you can really do to improve poor quality mp3, really just have to replace them with better quality mp3s or FLAC.

 

[PurpleAngel]

What is the advantage of converting non-DSD file to DSD during playback? Wouldn't the conversion introduce noise/artifacts?

Up scaling/converting your mp3s, to DSD will not improve audio quality, but imperfections in the audio might (should?) be amplified (more noticeable).

 

[old tech]

I really apologize if this is a dead horse I'm picking at, but I've read this article recently:
https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

and have been trying to find more information about the type of DAC that will be suitable for me. I actually did not know much about DSD before reading that page (searching for DSD is how I found it), and it might well be a biased article, since it is after all advertising their own product at the end. But anyway, I did take away a point the article made--I think, unless I misunderstood:

DACs that use native DSD decoding are best with DSD encoded source files. Such DACs process PCM encoded files in some inferior way due to conversions. At least that's what I've gathered--is that right? Now the question is, which DACs have good enough processors that have the more common nowadays native PCM encoding in high quality and are budget-friendly.

I actually have never looked at my music collection to see if PCM or DSD encoding is prevalent, and a lot of my library is mp3s shared by friends over the years, which is sub-optimal quality already. Based on what the article says, I gather that the percent of properly encoded DSD source material I have is about zero. If I do have some CDs that either have reasonable quality DSD encoding or can be converted to DSD, it still will be a small percent of my collection.

Should I then basically stay away from good budget DAC/amps such as Topping NX4 DSD which has native DSD decoding chip (XMOS XU208)?
http://www.tpdz.net/en/products/nx4dsd/index.htm
There's a non-DSD older version of this DAC/amp, but no word on what DAC processor is used, whether it's also a native DSD decoder.

Thanks in advance for any explanations and advice.

Reading through that article there are some problems with his statements around PCM. He focuses on quantisation errors and it's distortions but ignores dither which is part of the PCM equation. Claims such as "this is why higher resolution files sound only slightly better" on playback [than 16/44] is presented without any objective evidence and in ignorance of the some 30 years of controlled testing being unable to conclude there is any discernible difference.

 

[castleofargh]

This is precisely what's confusing me. If the original source files are usually NOT DSD or else are converted DSD just for the purpose of being played through a DSD DAC, what's the point of getting DACs with native DSD processing chips?

none.

Ok, so by this logic, all mobile DACs do native DSD if they are specced to process DSD? (I'm probably missing some other native formats, so that's probably incorrect reasoning). So those that are not specced to decode DSD don't have native DSD decoding chips but rather most likely PCM? I wonder if I want the latter for the near future if I won't be using DSD anyway. Or if I should ignore DSD/non-DSD decoding specs and just go for other parameters of a DAC/amp, such as noise, amp power, reviews of "soundstage" etc.

I believe you got confused by amirm's answer. mobile gears and most others are using delta sigma DACs, which "speak" PCM, not DSD. but the way they work is sort of a mix between pulse modulation(like DSD), and multibit PCM. that's a matter of inner working and you really don't have to care. also those DACs expect you to feed them PCM audio, not DSD.
I need to insist on this again, DSD can be seen as a niche compared to PCM audio, and same thing for the DACs able to play DSD. from the get go a DSD file is very heavy and has IMO nothing to do on portable gears.

reading your posts you seem to assume that we have those 2 typical ways of dealing with audio. that is not at all how it is. PCM is overwhelmingly used to contain music, and if you specify nothing while talking to people, everybody will assume that when you say DAC, you mean one that will read and convert PCM to analog. DSD is a failed format that almost died years ago and might still do just that soon enough.
you have no DSD in your library, so unless you're really curious, my personal advice would be to forget all about DSD right now. as I said, multichannel audio might be a reason why you would be interested. a few exclusive masters could be another reason. but you'll have to take my word for it, don't go wasting money on a device and purchase expensive SACDs for some misplaced expectations that somehow the signal is better or that the DAC does something more adequately. those are bad reasons to care about DSD. plus most DSD albums released to this day were initially PCM at some point and were converted to DSD. nobody likes to run in circle.
for now, and as far as I know for many years to come, PCM is and will be the audio standard.

 

[bigshot]

There are very compact mobile DAC/amps that have "native" DSD decoding, which is a reason I started this thread.

What are you using for a player? And what kind of headphones? I use an iPhone and Oppo PM-1s. I have no need of a DAC or amp with those.

 

[TidalWave]

I'm using a budget Motorola android-based phone (which I prefer to use as opposed to a DAP) and Audeze Sine + V-Moda M-100. The Sine headphones like a lot of input power, so initially I was going to get just an amp + use my phone. But I'm exploring the option of a DAC/amp combo, since the ones I'm looking at are in the same price range and approximate size as pure amplifiers.

 

[bigshot]

I'm not familiar with that particular phone. You should check into its line out. If you need a headphone amp, one without a DAC would be cheaper and a lot simpler. I'm guessing your phone doesn't have a lot of storage. I wouldn't give any priority to DSD playback. The file sizes for DSD are humongous compared to more practical formats, and the sound quality is no better than any audibly transparent codec.

 

[gregorio]

but I've read this article recently: https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

Others have pointed out various problems/incorrect statements in that article. You need to be aware though that we're not talking about an article that has a few faults/problems and we're just nit-picking those problems. The article is absolutely riddled with faults and even when the author does make a valid point, much of the time his explanation of those valid points is based on incorrect information/assumption. I don't intend to go through every incorrect statement and explain why it's incorrect, that would take a number of pages but there's one I will mention, as it underpins much of his article and also seems to have affected your thinking, and that's his constant use of the term "resolution":

If you remember nothing else, remember this: Within the audible range (20Hz - 20kHz), there is NO resolution difference between using a 44.1kS/s sample rate or any higher sample rate! In other words, between 20Hz and 20kHz sample rates of 88.2kS/s, 96, 192, 384 or even 2.822mS/s provides exactly the same resolution as 44.1kS/s. So how come high sample rates are called "high resolution"? The answer is essentially marketing BS! However, we can't simply call "hi-rez" an outright lie because in a sense, it's true. To explain this, let's take for example a sample rate of 88.2kS/s. With 44.1kS/s, when we reach an audio frequency of 22.05kHz we have no/zero resolution. With 88.2 we have the same resolution as 44.1kS/s within the audible range but instead of that resolution ending at 22.05kHz, it extends to an audio frequency of 44.1kHz. So, in a sense you could say that 88.2kS/s has double the resolution of 44.1kS/s. However that's very misleading because what you've really got is exactly the same resolution but covering double the frequency range. Likewise for 176.4kS/s but covering quadruple the frequency range (audio frequencies up to 88.2kHz). But of course, you can't hear above 20kHz and therefore you can't hear any of this additional/extended frequency range "resolution".

Because of this, not only is much of the article incorrect but so are some of your statements/questions, for example:

[1] It seems like perhaps improved formats of PCM are more the future rather than DSD ...
[2] DSD, however, is still used as an archival format for studio applications, and it's seen as a possible replacement for analog tapes.

1. How are the new PCM formats "improved"? Are they improved because they provide audio recording and reproduction in frequency bands which are many times beyond the limits of human hearing? Or, are they improved in terms of marketing, the marketing implication that you're getting 2, 4, 8 or however many times more "resolution"?

2. Not really related to any of the above but I'm not sure where this statement has come from. Sure, if something has been recorded in DSD you might as well archive it in DSD but I know of no instances of it being used as an archival format for anything other than original DSD recordings and, it would make no logical sense to do so. Of course, that doesn't necessarily mean there are absolutely no examples of this occurring but it would be a rare exception and certainly NOT common studio practice, as implied.

As castleofargh basically stated, the only reason why you might want to have the capability to play DSD is those instances where a particular master may only be available in DSD format and even then, I believe consumer software is available to convert to PCM with no audible loss of resolution/quality.

G