Does DSD sound better than PCM?

[solderdude]

I can see why that may be confusing to YOU.

The design of that filter is their task.
Not your responsibility to design an upsampler so they won't have to filter anything.
Afterall the majority of people will not be using any of your products and so they will have to make such a filter anyway.
Don't get me wrong, I applaud your efforts and like/agree most parts of your reasoning.
For me it's not needed though.

A personal question... does DSD sound better than PCM to you ?
I realize this is a tricky question because RB obviously is not the same as 192/24 for instance both PCM, and DSD64 is not the same as DSD256.
But in general... even if it only matters to you.

 

[March Audio]

Why don’t you make 16/44 DACs? All else is waste, isn’t it?

Well there still seems to be little proof that people can genuinely hear the difference between red book and hi red.

 

[Veri]

Well there still seems to be little proof that people can genuinely hear the difference between red book and hi red.

If all DACs used their bits overhead for volume control but dithered down to 16-bit I'm not certain people would even notice.. 16-bit/44Khz decoded (properly) to analog sounds absolutely fine and for most use cases it would be sufficient indeed.

But obviously, why not use 24-bit

 

[March Audio]

That highest signal consists partially of alias... With halfband filter, rejection at Nyquist is pretty much non-existent. So the band between 20 kHz and 22.05 kHz is mixture of base-band signal and alias of frequencies above it (22.05 - 24.1 kHz). In my castanets example -20 dB level hits this 22.05 - 24 .1 kHz band.



Hmmh, what radio starts there?

But yeah, there are radio transmitters even in ELF range for communicating with submerged submarines:
https://en.wikipedia.org/wiki/Extremely_low_frequency

And what level are the castenets? We are back to the same point where in normal audio recordings the levels aren't high. So above the audible range, 22 to 24 kHz which is reduced towards the audible range.

 

[March Audio]

You didn't do any exhaustive search even, you quickly picked something you found from your own collection. That proves really nothing.

And that doesn't have anything to do with in band aliases.



And I'll ask you to show some data about non-audibility. I assume perfect audibility unless proven in statistically realiable way otherwise. And it must be statistically reliable against entire population on earth.



His data showed one case for one ADC and then you extrapolate from that all-encompassing generalization that it covers all the cases. And you think that audible band starts exactly at 20000 Hz, so 20500 Hz is already totally inaudible.

Well I can post dozens of files if you like.

There is a wealth of data about the limits of human hearing. The classic 20kHz is stated for a reason. It's especially the case when we start talking about levels that are circa - 90dB where these aliases may be.

Regarding other ADCs I have a few here I can test.

 

[Frank Dernie]

I followed long threads on Computer Audiophile about HQPlayer several years ago.
Very impressive piece of work but since I don’t “do” computer audio, and back then it was PC only and I don’t “do” Microsoft (or Google while I am complaining) it was of only academic interest, hats off to @Miska though.
Personally I have satisfied myself that for my old ears if there is a difference between 44/16 and higher res I either fail to hear it at all (usually) or suspect (rarely) I may be able to hear a tiny difference but it is inconsequential to my musical enjoyment.
Also, given I listen in a room with a background noise around 30dB on speakers with probably worse than -60dB of distortion I, personally, see no point in any component with massively lower distortion than that or higher dynamic range than 16-bit.
I have been recording music for almost 60 years now, and whilst I can appreciate that an ADC with a real 21 bit dynamic range could record anything audible without needing a level control why would any sane person believe there was any point in a device to do so?
I can record at 192/24 but use 96/24 but only because I can.
16 bit is plenty both IME and IMO, DSD files are mahoosive for no point at all

 

[Miska]

And what level are the castenets? We are back to the same point where in normal audio recordings the levels aren't high. So above the audible range, 22 to 24 kHz which is reduced towards the audible range.

As I said, in the 22.05 - 24.1 kHz range level is -20 dB. Checked my maracas samples too and the relative level difference is the same.

 

[Hugo9000]

Oh, that reminds me of my favorite album:

Maracas ex Machina, by The Castanets Consort

you can imagine the difficulty of achieving anything resembling realistic playback!

 

[svart-hvitt]

Isn't this hostility against a new member conspicuous? One vocal manufacturer and several members cannot hide their repugnance for @Miska . Normally, @amirm welcomes new members and developers. But not this time; total silence. Prior to focusing on HQ Player, @Miska worked 7 years for Nokia, 7 years for Intel and 7 years for Image Soft. Conventionally, there's respect for experience. Why not here?

@Miska is criticized for perfectioning digital conversion. Almost a century ago (!), in 1937, Alec Reeves thought out PCM. His idea, dated back to 1937, initiated people's imagination that perfection of digital could represent analog. This idea was so strong, we're still dealing with PCM in 2019. And @Miska is dealing with perfectioning digital conversions at an amazingly high mathematical level.

@Miska is criticized for dealing with filters beyond red book standard, 16/44. Yet, nobody questions aging @amirm 's ability to distinguish 16/44 from hi-res (what does that tell us about younger listeners' ability?). And @Miska pointed to a refereed paper that indicates people can really hear beyond 20 kHz, up to 28 kHz. Is it written in stone that 16/44 is all we need? Could future research - for example relating to listening fatigue, "slow" and trained listening - open up the possibility that we need to revise our understanding of 16/44 versus hi-res? Is 16/44 truth?

If we were to practice the critique against @Miska on a general level, we'd need to criticize manufacturers of audio products that go beyond 16/44 and what's generally considered transparent. But we don't do that, do we? No, normally we applaud manufacturers of products that are more than what's currently considered transparent.

@Miska 's HQ Player does a lot of things. Conversion is one of them. At a price of €150 for consumers. So why this hostility against conversion perfection in this particular case?

 

[March Audio]

View attachment 22532

As I said, in the 22.05 - 24.1 kHz range level is -20 dB. Checked my maracas samples too and the relative level difference is the same.

So this is a test recording. Not an actual music recording. What level would they be in an actual music recording? Certainly not up to 0dB

 

[March Audio]

Isn't this hostility against a new member conspicuous? One vocal manufacturer and several members cannot hide their repugnance for @Miska . Normally, @amirm welcomes new members and developers. But not this time; total silence. Prior to focusing on HQ Player, @Miska worked 7 years for Nokia, 7 years for Intel and 7 years for Image Soft. Normally, there's respect for experience. Why not here?

@Miska is criticized for perfectioning digital conversion. Almost a century ago (!), in 1937, Alec Reeves thought out PCM. His idea, dated back to 1937, initiated people's imagination that perfection of digital could represent analog. This idea was so strong, we're still dealing with PCM in 2019. And @Miska is dealing with perfectioning digital conversions at an amazingly high mathematical level.

@Miska is criticized for dealing with filters beyond red book standard, 16/44. Yet, nobody questions aging @amirm 's ability to distinguish 16/44 from hi-res (what does that tell us about younger listeners' ability?). And @Miska pointed to a refereed paper that indicates people can really hear beyond 20 kHz, up to 28 kHz. Is it written in stone that 16/44 is all we need? Could future research - for example relating to listening fatigue, "slow" and trained listening - open up the possibility that we need to revise our understanding of 16/44 versus hi-res? Is 16/44 truth?

If we were to practice the critique against @Miska on a general level, we'd need to criticize manufacturers of audio products that go beyond 16/44 and what's generally considered transparent. But we don't do that, do we? No, normally we applaud manufacturers of products that are more than what's currently considered transparent.

@Miska 's HQ Player does a lot of things. Conversion is one of them. At a price of €150 for consumers. So why this hostility against conversion perfection in this particular case?

It's not hostility. These are genuine technical questions which aren't unrelated to the idea of dsd's advantages.

What's conspicuous is you repeatedly making these interventions. Are you still irritated by me asking you not to disrupt another thread with totally off topic talk about your political views of product sustainability?

That research was somewhat unconvincing and would need to be repeatable by other tests. 110dB 50cm from the ear, representative of any realistic signal no.. At the very best that demonstrates just how insensitive the ear is if it can indeed perceive those frequencies assuming there is no other issue or clue in the signal. The results do not tally with decades of other research.

We could make a file of a 20kHz tone at say - 50dB. You set your system at a reasonably loud listening level with music and then play the tone and see if you can hear it.

 

[March Audio]

It an Aura. Which is pretty much the M3 in better build quality. And yes depending on the settings on the interface it is around .75 ohm at the upper frequencies. Here is a plot from long ago of the A1, but it is pretty much the same as mine in impedance.

View attachment 22511

That is a challenging speaker. I suppose the saving grace is that the impedance low is at 20kHz where signal levels are very low.

 

[Purité Audio]

Invent a problem then sell you the ‘solution’.
Keith

 

[svart-hvitt]

Why?

It's not hostility. These are genuine technical questions which Rent unrelated to the idea of dsd's advantages.

That research was somewhat unconvincing and would need to be repeatable by other tests. 110dB 50cm from the ear, representative of any realistic signal no.. The results do not tally with decades of other research.

We could make a file of a 20kHz tone at say - 50dB. You set your system at a reasonably loud listening level with music and then play the tone and see if you can hear it.

You just wave off peer-reviewed research that doesn’t fit your view.

However, I also said @amirm can hear beyond 16/44. And he’s aging.

It’s true that subjective testing has shown that going beyond 24/48 is a difficult position to argue for. But the same kind of subjective tests have been used to promote lossy formats. These lossy formats have later been found to represent problems, artifacts.

I don’t understand your opposition to a perfectionist engineering attitude.

 

[March Audio]

You just wave off peer-reviwed research that doesn’t fit your view.

However, I also said @amirm can hear beyond 16/44. And he’s aging.

It’s true that subjective testing has shown that going beyond 24/48 is a difficult position to argue for. But the same kind of subjective test have been used to promote lossy formats. These lossy formats have later been found to represent problems, artifacts.

I don’t understand your opposition to a perfectionist engineering attitude.

No, I questioned research that doesn't tally with other data. I questioned the validity of blasting someone at 110dB directly into the ear and the potential for this to produce other tell tales.

Amir is the expert on the subject of lossy formats, but Im not aware of there ever being any concensus that lossy formats are devoid of issues. Quite the opposite in fact, there has always been concern over them.

I have no objection to it. I'm framing this within the constraints of the real world. There is so much in this hifi area that is questionable as to its actual audibility that its worth questioning and challenging.

DSD is one of those things and the conversation has led to a wider but not totally unrelated.discussion

 

[svart-hvitt]

No, I questioned research that doesn't tally with other data. I questioned the validity of blasting someone at 110dB and the potential for this to produce other tell tales.

Amir is the expert on that subject, but Im not aware of there being any concensus that lossy formats are devoid of issues.

I have no objection to it. I'm framing this within the constraints of the real world. There is so much in this hifi area that is questionable as to its actual audibility that its worth questioning and challenging.

DSD is one of those things and the conversation has led to a wider but not totally unrelated.discussion

@March Audio , you wrote:

«There is so much in this hifi area that is questionable as to its actual audibility that its worth questioning and challenging».

Still, I bet a beer that you’ll market a new and improved dac when time is due.

Or do you think you’ve made your last dac?

You should use the same principles when looking at yourself as when looking at others.

 

[March Audio]

Invent a problem then sell you the ‘solution’.
Keith

To be fair I don't think that is the case here, but the technical issues do need to be put in context and realistically judged for any real world benefit.

 

[March Audio]

@March Audio , you wrote:

«There is so much in this hifi area that is questionable as to its actual audibility that its worth questioning and challenging».

Still, I bet a beer that you’ll market a new and improved dac when time is due.

Or do you think you’ve made your last dac?

You should use the same principles when looking at yourself as when looking at others.

I will indeed. Part of the reason I am in this is to make money. I make no apology for that.

I will make a balanced output dac because that has specific advantages regarding interconnection signal integrity. I know it will marginally technically outperform the current dac, but you won't see unjustifiable marketing bull.

I will spend energy on other areas such as dsp active speakers where there are genuinely large performance gains to be had and not on dac technology which has clearly come close to its limits.

 

[RayDunzl]

However, I also said @amirm can hear beyond 16/44. And he’s aging.

Uh, I would interpret his posts as revealing he can hear faults within 16/44, not beyond.

 

[Miska]

I will spend energy on other areas such as dsp active speakers where there are genuinely large performance gains to be had and not on dac technology which has clearly come close to its limits.

Interestingly people are running multi-way cross-overs with digital room correction using my software...

I just like to combine all DSP into one powerful place, no need to scatter it all over in small pieces.

So this is a test recording. Not an actual music recording.

It is actual music I'm playing myself. Or are you trying to say that my music is not worthy of consideration because it doesn't fit into your expected frequency envelope? Sounds a lot like MQA... I'm the only one in the world who has recorded castanets, maracas or soprano glockenspiel playing solo? Nobody else could of course do anything like that or if they do it is unimportant?

We need to get some kind of approval for content before it can be released out there so that it doesn't break expected rules?

but Im not aware of there ever being any concensus that lossy formats are devoid of issues. Quite the opposite in fact, there has always been concern over them.

Interestingly we've been talking about errors of about similar level... And they appear predominantly in the top octave. That sounds a bit selective...