DSD: Explain it to me because I'm not getting it

[blownsi]

As I see it, the current trend is to buy a dac that doesn't play your audio files. Instead you must buy a hot rod of a computer to convert your files into a format that this new dac can play. This is supposed to yield better results? How? At the end of the day aren't you just playing back what you started with?

[ericuco]

There are lots more people on this site that are more technically knowledgeable than me but here is my take:

 

Several years ago, the big issue was getting "bit perfect" data to the DAC. However, it was soon realized that the various DAC chips are internally monkeying with the data, converting it to whatever format it needs.

 

So, the current trend is to let the computer, which has more resources than the DAC chip, do the conversion so that the DAC chip gets the format it is wanting. This lessens the load on the DAC chip. It also allows you to use more advanced conversion algorithms with the computer (vs DAC chip). If you look at HQ Player which has a number of algorithms (filters, etc.), the computer demand to do the conversion can be considerable especially if doing PCM > DSD.

 

Hope this helps.

[YashN]

Having the computer do the heavy processing (if you choose to do so), allows the DAC to do less when doing the D/A => better SQ in most DACs.

 

Now, as for format and implementation, let's say you have a DAC which does DSD best, then you'd rather send it pure DSD. You could even have better results using RedBook PCM -> DSD conversion prior to sending the data to it.

 

It's a matter of preference and implementation, so it's best to try the various settings rather than think about it or discuss it.

 

Get set up and listen to various ways of playing back.

[EuroDriver]

I would make three points about the trend towards DSD and computers doing the playback.

 

DAC's need filters, the filters for DSD are at much higher frequencies than those for PCM and can have more gentle slopes which are less damaging to sound quality than filters with steep slopes

 

there is a SQ gain to be had converting PCM to DSD, but to do it well is computationally intensive. A i7 or similar computer has an order of magnitude more math computing power than what is available in hardware at reasonable cost, and it's much easier to update / improve software than it is to update firmware. However the PS Audio DirectStream DAC does a very good job in hardware and firmware. Playback design also does a good job in hardware / firmware.

 

In software, HQ Player does an amazing job, but there is a learning curve, and the PC hardware, either one or two computers, need to be fit for purpose

 

the last point is doing computational intensive processing creates quite a bit of RF. Doing this computation in a separate chassis isolated by USB or even better 2 PC's and Ethernet from the DAC makes for a better sounding system

[kilroy]

All else equal, DSD native or upsampled usually sounds better.

[blownsi]

Thank you all very much for the explanations.

[trappy]

All else equal, DSD native or upsampled usually sounds better.

 

Not in my system it doesn't. And I have a DAC regarded as a DSD champion.

[EuroDriver]

Not in my system it doesn't. And I have a DAC regarded as a DSD champion.

 

You need to try out HQ Player ;-)

or Bootcamp Windows and try Foobar + SACD

 

I guess you find the volume control on the W4S mPre sounds more transparent than the volume control on the E22, can you confirm if this is the case ?

[trappy]

I've tried out HQP. I think I just don't like DSD all that much. The exa is awesome on PCM so I'm not troubled by it.

 

I find the preamp sorts out the musical strands and firms up the vocals a bit. It's subtle, but noticeable. On the other hand it's not quite as transparent at super low levels as going direct into the amp. It's kind of jaw-dropping how low you can go and still hear all the detail with the exa.

 

I prefer the headphone amp on the e22 to that of the mPre, though.

[kurb1980]

I very much like the DSD with HQPLAYER paired with the right DAC is key. Experimenting with the modulators and filter until you find the sweet spot is tedious but worth it. DSD has a smoothness and airiness very similar to analog. In the digital word it sounds the most like analog when it's done right!

[bmoura]

I very much like the DSD with HQPLAYER paired with the right DAC is key. Experimenting with the modulators and filter until you find the sweet spot is tedious but worth it. DSD has a smoothness and airiness very similar to analog. In the digital word it sounds the most like analog when it's done right!

 

The right DAC is one key. Another is the music downloads you get.

 

Downloads sourced from Analog Tape to DSD or DSD recorded will show off DSD's capabilities much better than a PCM to DSD transfer in my experience.

[Paul R]

I've tried out HQP. I think I just don't like DSD all that much. The exa is awesome on PCM so I'm not troubled by it.

 

I find the preamp sorts out the musical strands and firms up the vocals a bit. It's subtle, but noticeable. On the other hand it's not quite as transparent at super low levels as going direct into the amp. It's kind of jaw-dropping how low you can go and still hear all the detail with the exa.

 

I prefer the headphone amp on the e22 to that of the mPre, though.

 

Some people are like that, loving the PCM sound frothier DACs. I don't think they should change or worry about DSD.

 

Other folks like the sound of DSD much much better.

 

Personal Preference?

[bmoura]

Some people are like that, loving the PCM sound frothier DACs. I don't think they should change or worry about DSD.

 

Other folks like the sound of DSD much much better.

 

Personal Preference?

 

Agreed. Having more musical choices is good.

[firedog]

Some DACs function better with either DSD or PCM. Some people prefer the sound of PCM, some prefer DSD. Hence conversion before sending signal to DAC.

[Kelly]

One thing to consider is the versatility of PCM vs DSD when buying music: there is a wealth of equalization and other DSP that is available for PCM information vs almost nothing for DSD data. Room correction, digital crossover, etc can make very tangible improvements as opposed to the (to some people) debatable merits of the intrinsic file formats. Furthermore, if you would someday like to have music spread around the house with a system like Sonos, Bluesound or any of the other networked/wireless music systems then your DSD will need to be converted to PCM and stored before it is useful.

 

So I guess there are two facets to the discussion:

 

1.) what data does a particular dac handle better than another

2.) which format do I want to buy my music in? am I happy to convert when needed or do I want to stay more "pure"

[Superdad]

Some people are like that, loving the PCM sound frothier DACs.

 

Paul: I hear a lot of things in experimenting with PCM and DSD DACs, but neither has ever given me a more or less frothy sound. I'm just not a latte guy--I like my coffee and my audio backgrounds black!

[Solstice380]

Some people are like that, loving the PCM sound frothier DACs. I don't think they should change or worry about DSD.

 

Other folks like the sound of DSD much much better.

 

Personal Preference?

 

Just look at the number of requests for PS Audio to "voice" versions of the Direct Stream DAC firmware!

[Solstice380]

Paul: I hear a lot of things in experimenting with PCM and DSD DACs, but neither has ever given me a more or less frothy sound. I'm just not a latte guy--I like my coffee and my audio backgrounds black!

 

+1

[Paul R]

Paul: I hear a lot of things in experimenting with PCM and DSD DACs, but neither has ever given me a more or less frothy sound. I'm just not a latte guy--I like my coffee and my audio backgrounds black!

 

Dang this iPad!!!!!! And double dang on the built in spelling checker!

 

Mmm- it's Quantum Foam!! Yeah, that is what it is....

[trappy]

Some people are like that, loving the PCM sound frothier DACs. I don't think they should change or worry about DSD.

 

Other folks like the sound of DSD much much better.

 

Personal Preference?

 

I have enjoyed some native DSD and I'll probably get some more. But in general, I find I like the airiness for a few minutes, but then I find it soft and weirdly glossy and it starts to annoy me. This is is particularly so with PCM->DSD conversion, which I don't like at all. Not in JRiver, not in HQP, ASIO, or Core. The space gets bigger and airier, but it sounds all wrong. Offstage instruments in classical pieces sound particularly wrong.

 

Chacun a son gout.

[EuroDriver]

The space gets bigger and airier, but it sounds all wrong. Offstage instruments in classical pieces sound particularly wrong.

 

Chacun a son gout.

 

Does this effect on the sound stage show it self more with the headphones than with the speakers, or it's even worse with the speakers ?

 

What sort of room treatment do you have ?

[Miska]

I've tried out HQP. I think I just don't like DSD all that much. The exa is awesome on PCM so I'm not troubled by it.

 

It's more down to which oversampling filters and delta-sigma modulators you like. exa is based on ESS Sabre which is delta-sigma design and thus PCM gets internally converted to DSD-like bit stream before hitting the actual D/A conversion stage.

 

So you can choose either to use DSP algorithms inside ESS Sabre, or do it externally and feed DSD there and as a result get much less DSP processing inside the DAC chip (but still some).

 

I have e28 here and I always feed it at DSD256, never anything else.

[audiventory]

If I don't mistaken, native PCM (stair voltage) DACs currently unavailable (or almost unavailable) on consumer market, isn't it?

[Skeptic]

DAC's need filters, the filters for DSD are at much higher frequencies than those for PCM and can have more gentle slopes which are less damaging to sound quality than filters with steep slopes

But haven't your PCM files already been filtered?

I can see upsampling first if you want to do additional processing, so that has less effect on the original audio, but I'm not really sure that there are any filtering benefits for files which are already in a PCM format.

 

I can see someone perhaps wanting to upsample if their DAC uses linear-phase upsampling and you want to use minimum-phase upsampling.

Going to the highest rate that the DAC supports will probably help minimize any pre-ringing for example.

 

But I feel like most people that prefer DSD like it because it colors the sound.

With PCM you can do a lot of processing without affecting how the file sounds, because it's a multi-bit format.

Because DSD is a 1-bit format, there are a lot of compromises to be made so each modulator may have its own type of "sound".

 

I have enjoyed some native DSD and I'll probably get some more. But in general, I find I like the airiness for a few minutes, but then I find it soft and weirdly glossy and it starts to annoy me. This is is particularly so with PCM->DSD conversion, which I don't like at all. Not in JRiver, not in HQP, ASIO, or Core. The space gets bigger and airier, but it sounds all wrong. Offstage instruments in classical pieces sound particularly wrong.

I don't think it is surprising really. Classical music fans were quick to abandon vinyl in favor of CDs, while some people stuck with it for rock or acoustic recordings due to the "sweeter" sound.

If people consider DSD to be "more analog" as in "more like vinyl" then I'd expect classical music fans to stick with PCM.

 

It's more down to which oversampling filters and delta-sigma modulators you like. exa is based on ESS Sabre which is delta-sigma design and thus PCM gets internally converted to DSD-like bit stream before hitting the actual D/A conversion stage.

 

So you can choose either to use DSP algorithms inside ESS Sabre, or do it externally and feed DSD there and as a result get much less DSP processing inside the DAC chip (but still some).

 

I have e28 here and I always feed it at DSD256, never anything else.

I hear this argument a lot, and obviously you are very knowledgeable on the subject, but I'd much rather send it a multi-bit input than convert everything to 1-bit first.

While some people may not consider them to be "multi-bit" DACs, they are not 1-bit DACs either, and do all of their internal processing in a multi-bit format.

So while you may prevent the DAC from doing any internal upsampling, it's going to go through remodulation which is more likely to have a detrimental effect on the sound, in my opinion.

[PeterSt]

I have enjoyed some native DSD and I'll probably get some more. But in general, I find I like the airiness for a few minutes, but then I find it soft and weirdly glossy and it starts to annoy me. This is is particularly so with PCM->DSD conversion, which I don't like at all. Not in JRiver, not in HQP, ASIO, or Core. The space gets bigger and airier, but it sounds all wrong. Offstage instruments in classical pieces sound particularly wrong.

 

Chacun a son gout.

 

A bit pretentious perhaps ... but this is about the ringing implied by the filtering.

Going into the technical details (of that filtering) may be not the most easy way to explain the exhibit, so I will try to explain it by a complete different means. From there "we" might proceed :

 

Let's first state that the DAC I use is the most accurate beast. Mind you, this "accuracy" is implied by the non-ringing in the first place (no pre-ringing, no post ringing). Oh ? he starts about the filtering after all ? no, not really, but it easily shows what happens when ringing of any kind is added. Could try to explain that via filtering, but more easy will it be to explain it by euphonic "behavior" which is more easy to recognize, I hope. So I start out with a setting which sounds very accurate to me. Nice ...

 

What this means is that the tiniest sounds are as accurately boundaried, say, as they are. A small bell is just that (and not too small either if all is right) and a larger voice is that just the same.

Now I am going to apply "footer" means, that give me a super wide (and deep) stage. Superrr !

Yes, for a day. Because what happens ? with the "spreading" of the stage, all spreads. That small bell too. It remains a bell sound, but it smears over a wider portion of the stage than should. In your own words "sounds all wrong". It's a gag.

 

What I do with the footers is implying a reactance between the DAC and the music itself. How ? well, I decouple it from its stiff base too much. And when that is done, the music "moves" the D/A converter. Say the fragile oscillator in there to begin with. Very very roughly put, the music is originally directed to the left, while the movement goes opposite because some later (reacts to the impacting sound wave itself). Now there's an oscillating sibilance.

Btw, you will only be able to understand this somewhat when you know that footers matter. And keep in mind : they matters the most when you are theoretically able to create that very accurate base the first.

 

Briefly into the ringing filters : When an originally "one sample" transient is now smeared over more samples (filtering is working like that) then this "echo" runs into the next sample. If this is repeatedly happening (and it is) then this can all be seen as a similar oscillation as I just described. So it is not only that this one transient (meant to form a small bell in space) is smeared somewhat so the bell already becomes larger because of that, but the next part of the bell is deformed also. It becomes a bit messy.

 

Nice stupid theory perhaps, if you only know that any ringing filter for me works out very similar as "wrong footers". They buzz. An euphonic sound to begin with. Zzzzzooms into space. Also makes the sound move while this originally is not so.

 

Now we can strart the debates on how DSD is far better in the ringing department than PCM for the 1000th time (and please do so if deemed necessary), but no-ringing is still that and this is what I base upon my just given real life perception of it.

 

Peter