The Best DAC is no DAC

Yeah I know. People always hear an improvement, no matter what ridiculous changes they make. Black plastic discs also cause audible improvements, my wife heard it from the kitchen...
If you want to discuss at that level then no thanks.

Converting PCM to DSD basically just adds noise, and lots of it. That's just a fact.
So this noise, that masks low level high frequency content, has to be suppressed with a filter. So you're already using multiple filters: one to upsample, one to shape the insane amounts of quantization noise, one to suppress that noise. The last might be a filter in the DAC to prevent damage of downstream equipment...
So you very likely get lots of phase shift. There are more problems with DSD that have been identified years ago, but of course that doesn't stop people from making ridiculous claims. 🙄

Of course you can get the same with PCM, without the noise, with less filters and without inefficient algorithms that stress the CPU.
 
Last edited:
Yeah I know. People always hear an improvement, no matter what ridiculous changes they make. Black plastic discs also cause audible improvements, my wife heard it from the kitchen...
If you want to discuss at that level then no thanks.

Converting PCM to DSD basically just adds noise, and lots of it. That's just a fact.
So this noise, that masks low level high frequency content, has to be suppressed with a filter. So you're already using multiple filters: one to upsample, one to shape the insane amounts of quantization noise, one to suppress that noise. The last might be a filter in the DAC to prevent damage of downstream equipment...
So you very likely get lots of phase shift. There are more problems with DSD that have been identified years ago, but of course that doesn't stop people from making ridiculous claims. 🙄

Of course you can get the same with PCM, without the noise and without inefficient algorithms that stress the CPU.

So you have no experience of this topic and we're just being ridiculous; fair enough.
 
So you have no experience of this topic and we're just being ridiculous; fair enough.
I see you have to continue the discussion with insults and on a naive and childish level because you cannot deal with facts and have no arguments.

I expected nothing else. It's always the same... I could be also be talking to a clairvoyant psychic who reads tea leaves here, no difference in response. 😀
 
Last edited:
I see you have to continue the discussion with insults and on a naive and childish level because you cannot deal with facts and have no arguments.

I expected nothing else. It's always the same... I could be also be talking to a clairvoyant psychic who reads tea leaves here, no difference in response. 😀


Clearly someone who has no idea how modern SDM DAC chips work, or chipless DSD DAC's for that matter.
 
Clearly someone who has no idea how modern SDM DAC chips work, or chipless DSD DAC's for that matter.
I don't know if nautiboy knows how they work, but what would be more important is an understanding of digital audio in general. DSD is a tradeoff. By converting PCM to DSD you get the worst of both worlds combined. xD
 
Last edited:
I don't know if nautiboy knows how they work, but what would be more important is an understanding of digital audio in general. DSD is a tradeoff. By converting PCM to DSD you get the worst of both worlds combined. xD


I was referring to you. Please explain how a modern SDM DAC chip like the ESS 9018 handles PCM audio?
 
Last edited:
Did you even read what I wrote? What's the argument for converting PCM to DSD other than adding noise (effectively destroying low-level high frequency content), phase shift, etc. and stressing your CPU?

Why should I explain a modern Delta-Sigma DAC? What's your point? Interpolation filters? Multibit architectures?
Please try to make a coherent argument...
 
Last edited:
Did you even read what I wrote?

Why should I explain a modern Delta-Sigma DAC? What's your point? Interpolation filters?
Please try to make a coherent argument...

It all ends up the same anyways. Only when you feed these chips PCM, the data has to go through the SDM/SRC section of the chip. Not the case when you feed these chips DSD.DSD bypasses this section. By handling the PCM- SDM/SRC on the PC end with HQplayer, you can take advantage of much better algorithms than the resource constrained chips can process.

Better yet is to use DAC chips that allow bypassing of the SDM/SDM in the chip all together. A great example is the AKM AK4490. Even better is chipless, if you can get all the filters right.
 
Last edited:
It all ends up the same anyways.
Unless you mean an analog signal at the output, no it doesn't.

Only when you feed these chips PCM, the data has to go through the SDM/SRC section of the chip.
Which are advanced multibit delta-sigma architectures for years now because that way we could achieve much cleaner output than 1-bit (DSD). Some chips even convert DSD back to PCM...
Also, the interpolation filter is typically configurable (including using your own) in high-end DACs and is effectively bypassed by resampling to a higher rate on the computer. (No, that's not the same what you're doing.)

Not the case when you feed these chips DSD.
You can also fill lawnmower oil into a Ferrari. 😛

By handling the PCM- SDM/SRC on the PC end with HQplayer, you can take advantage of much better algorithms than the resource constrained chips can process.
No, you force quantization to 1 bit so you need extreme rates and noise shaping such that you do not arrive at significantly worse performance than even 96/24.

Even better is chipless, if you can get all the filters right.
And you have measurements that back up this claim, right?
 
Last edited:
I don't know if nautiboy knows how they work, but what would be more important is an understanding of digital audio in general. DSD is a tradeoff. By converting PCM to DSD you get the worst of both worlds combined. xD

I don't think this is the place to resume the PCM vs DSD permathread that pops up everywhere. No one is being forced to do this project. No one is being coerced to convert to DSD.

Assuming you *are* interested in DSD and/or PCM -> DSD and assuming you want to build something discrete and as simple as possible (but no simpler 🙂 then that is what is being discussed here.
 
So you don't like what we are doing. Your entitled to that opinion/point of view. You've made your point.

Thanks.
Thanks for the respectful response, but it's not about my opinion. Most of what I wrote does not depend on opinion, as opposed to what appears to be the only thing that some people cling on to here, which is why I get poor responses like this:

His mind is frozen at around 2004-2005. No point in arguing, it's wasting energy.
I see that you have nothing intelligent or factual to say in response to what I wrote. I rest my case.
 
Assuming you want to upsample to DSD256, you will need a fair amount of processing grunt. I use an HP Z210 Workstation with Xeon E3-1240 3.3GHz Quad Core processor and that does the job nicely. That processor benchmarks at just a tad under 8000 so I would say you should aim for something similar or better. You can find benchmarks for suitable processors here;

https://www.cpubenchmark.net/high_end_cpus.html

BTW, because I wasn't sure what this DSD journey would bring I bought my workstation as 'previously loved' for only £200; worth considering if you're on a shoestring or aren't sure if this is for you.

The DSD filter board I'm sending you is designed to mount on the header pins of the JLSounds USB board so, if you're planning to use a WaveIO you'll have to work out how to hook it up, keeping the data connections nice and short.

Another BTW, if anyone is planning to purchase HQPlayer I recommend contacting Gianluca (Bibo01) who is an authorised reseller and will give you a better price than the HQplayer (signalyst) website.

Ray
Thanks. That seems like a real heavy duty spec to me. What cpu loads are you typically seeing on this server when converting to dsd?
 
I've been out for the evening...

I attended a Christmas themed concert of early(ish) music played on (mostly) reproduction medieval instruments - amazing organic sounds and not a sound system in sight. The acoustics of the venue weren't brilliant, there were a few minor mistakes and the hurdy-gurdy kept needing tuning but boy was it fun. I've never heard an audio system that gets anywhere close to recreating a live event but it's good to be reminded that the simple enjoyment of music is why I pursue this hobby.

Ray
 
ok converting from this to that it sounds bad but to my ears and my friend it is sounded almost like his diy record player, deem good one. or ignore us.
What you are hearing, if anything at all since I doubt you guys have done proper blind tests, is the conversion filters and phase shifts.
In some cases of inadequate filtering high amounts of high frequency noise could make it into downstream equipment and also cause audible differences (or damage).

@xnor, a quick search of your thread participation over the last few months shows you participate in order to argue.
That is a lie and a poor attempt at shooting the messenger. I'd prefer straight up insults instead of such dishonest behavior, if that is the best you can do..

This thread is a collaborative exploration of new techniques and ideas. Experiments rule and arguments are moot.
Great, I'm telling you why it's not such a great idea. And this is not based on some philosophical arguments or biased claims about subjective improvements but simple facts about how digital audio works.

Just compare the spectra of 192/24 PCM and DSD before the application of a noise suppression filter. Then overlay the spectrum of a DXD recording.

I'm sorry but DSD just isn't that great.