DAC types and their sonic signature

[Blumlein 88]

I did a similar test with a smaller local audiophile forum group, only it was original plus 5 generation copies, and the task was to find the original plus to find copying order for the rest. Few guys did good, one was almost perfect, nobody was 100%. The rest, so-so. I myself at the time was waiting for my speakers to be repaired so I only listened through cheapo headphones through laptop out. Difference was too small for me. Later the results were published so I never bothered to try it through my system.

But something else here...outside of this being an excercise, this blind test has no similarites with the usual audiophile situation (changing components in a hifi chain), so I'm not sure about the purpose.

Why is it not part of the normal audiophile situation? Some files are the original digital file. Another has been through 8 DA to AD conversions. If a DAC or ADC has a sonic signature then 8 times that much combined should be audible. The difference was difficult to perceive. While there certainly are some audible DACs, most measure very well, and this would make me think any difference is fleeting and very small.

 

[zalive]

Why is it not part of the normal audiophile situation? Some files are the original digital file. Another has been through 8 DA to AD conversions. If a DAC or ADC has a sonic signature then 8 times that much combined should be audible. The difference was difficult to perceive. While there certainly are some audible DACs, most measure very well, and this would make me think any difference is fleeting and very small.

In short, because in audiophile situation you change equipment in the system, while here various files altered by DA to AD and forth are being played through the same system.

 

[Blumlein 88]

In short, because in audiophile situation you change equipment in the system, while here various files altered by DA to AD and forth are being played through the same system.

Yes, so I don't see the problem. Changing equipment changes the signal reaching the speakers. Recording a change in equipment with decent accuracy will change the signal reaching the speakers. You'll hear the difference if there is one.

Maybe some interaction with other connected gear is different in other ways than just recording the other gear. Still the difference is of the same kind. If DACs have much of a signature, increasing the difference vs transparency by 8 times should highlight that even more. To argue otherwise would be to think changes in recordings like this mike instead of that one or this EQ rack instead of another is to have no effect that survives to be heard.

 

[zalive]

I don't see a problem with the test, either. However it's for the another piece in chain. Which you never change usually. It's a fine demonstration of a quality of DA-AD by itself, that's about it, the way I see it.

As for the test per se, it involves recognition and preferrence as well. So let's suppose you will hear a difference between each sample. Hearing is joy enough, you need to recognize the last gen copy. Recognition is based on presumption you'll prefer previous gen copies for fidelity. What if your personal preference might be different? With the mp3 I sometimes prefer codec compressed intervention to the original, if the original sounds so-so. So you see, even hearing a difference won't ensure which copy is last gen - to ensure this, each copy should sound 'worse' to your ear as well.

If you don't mind a colourful description, this test reminds me of hypothetical situation where the food is eaten then digested and thrown out through natural outlet, which is then eaten again. Then repeat the cycle 8 times in a row and the task is to recognize the final sh*t

 

[solderdude]

It's a fine demonstration of a quality of DA-AD by itself, that's about it.

Yep, that's the goal of the test. It includes the ADC and DAC chain 'errors' to multiply to kind of show the degradation in fidelity magnified.
Would be fun to repeat the test but recording the output of a filterless NOS DAC 8 times.

If you don't mind a colourful description, this test reminds me of hypothetical situation where the food is eaten then digested and thrown out through natural outlet, which is then eaten again. Then repeat the cycle 8 times in a row and the task is to recognize the final sh*t

Not entirely a valid comparison because after the first 'loop' the signal is indistiguishable from the original yet the excrement is quite distinguishable after the first 'passage'.
A more correct comparison would be to consume your excrement and do this 8 times in a row and see if you can taste a difference.
Even then, one may get accustomed to the taste and may not find the 8th time as 'bad' as the first time.

 

[Krunok]

Yep, that's the goal of the test. It includes the ADC and DAC chain 'errors' to multiply to kind of show the degradation in fidelity magnified.
Would be fun to repeat the test but recording the output of a filterless NOS DAC 8 times.



Not entirely a valid comparison because after the first 'loop' the signal is indistiguishable from the original yet the excrement is quite distinguishable after the first 'passage'.
A more correct comparison would be to consume your excrement and do this 8 times in a row and see if you can taste a difference.
Even then, one may get accustomed to the taste and may not find the 8th time as 'bad' as the first time.

We had some really weird analogies on this forum but with this "eat your shit" thing you guys have definitely brought it to the new level.

 

[solderdude]

There is a guy that consumes airplanes etc.
We could as him to eat some Schiit.

 

[Calexico]

I did the test of the files that went 8 times into adc dac and this is what i found:
- original files were sounding not pleasant
- the processed files were also not sounding pleasing

For this it was difficult to choose the best one.
I admit it was difficult to hear differences.

I have to say that the result of this test is dependent of the system.
This test just show that making multiple filtering doesn't change the sound comparing to filtering only one time.
But as nobody can compare with the high res studio master we cannot validate that one time filtering at 44.1khz sound as good as the master record with its root filter.

Also i have to say that ringing remain the same if you do 1 time or 8 time the conversion.
It's in the video you always link whatever question is asked

So i don't really see the point of this test.

 

[zalive]

Not entirely a valid comparison

It was just a joke

 

[solderdude]

But as nobody can compare with the high res studio master we cannot validate that one time filtering at 44.1khz sound as good as the master record with its root filter.

That wasn't the object of the test nor how pleasing the music is.
It merely shows the quality of an ADC/DAC loopback in the analog domain ... 8 times as 'bad' as the original 44.1 file.
So original file -> DAC -> analog -> ADC and the capture replayed by the same DAC and analog into an ADC again .... looped 8x
What the test shows is that even with a not high-end DAC the audible differences aren't that big (quite measurable though).

It could ease the mind of some people knowing they do not need to worry much when using the same DAC as used in the test when reproducing a 44.1 file. At least the minds of some.

 

[zalive]

Would be fun to repeat the test but recording the output of a filterless NOS DAC 8 times

We would need one which can do AD too to loop it. But with a good chip. Some like TDA1543 are cheap and technically inferior. OTOH AD1865 or TDA1541 or PCM63 are real deal.

I compared my AD1865 with MyDac when it comes to piano sound. Trying to find out why multibit reproduced piano sounded more real. The obvious thing I can discern is...dynamic. AD1865 delivers some impressive attack on piano. Well the power section has 45W, it's without any hint of compromise. But my audio designer friend says the realistic dynamics of AD1865 is actually superior to usual 1-bit chips. Tried to explain to me why so but I already had couple of beers

 

[solderdude]

The obvious thing I can discern is...dynamic. AD1865 delivers some impressive attack on piano. Well the power section has 45W, it's without any hint of compromise. But my audio designer friend says the realistic dynamics of AD1865 is actually superior to usual 1-bit chips. Tried to explain to me why so but I already had couple of beers

That should be very easy to confirm using captures of a piano recording and Paul's software.
I assume with 1-bit chips you mean Delta Sigma DAC chips. These aren't 1-bit also not when you had a couple of beers

 

[zalive]

This should be easy to confirm, yes. But why software? What's the purpose/function of Paul's software? Delta sigma, yes. 1-bit is communication within them during a conversion, no?

 

[hetzer]

This should be easy to confirm, yes. But why software? What's the purpose/function of Paul's software? Delta sigma, yes. 1-bit is communication to them, no?

Delta sigma architecture doesn't mean that it is 1bit. Modern audio DAC use 4~5bit architecture.

 

[zalive]

Delta sigma architecture doesn't mean that it is 1bit. Modern audio DAC use 4~5bit architecture.

I see. I remember some really old explanations from the early 90's when they used to make 1-bit architecture sigma-delta DACs.

 

[zalive]

Speaking of this, one local audio designer is currently developing a R2R 1-bit chipless DAC-preamp design. It will be interesting to hear it once he finishes it and delivers it to the market.

 

[solderdude]

This should be easy to confirm, yes. But why software? What's the purpose/function of Paul's software? Delta sigma, yes. 1-bit is communication within them during a conversion, no?

https://www.audiosciencereview.com/...test-deltawave-null-comparison-software.6633/

When one assumes there are audible differences between DACs then the actual analog waveforms MUST be different.
Pauls software compares 2 recordings and matches them in time and amplitude so that different waveforms can be compared.
To do this quite soffisticated software is needed that alligns and matches those recordings.
This is what the software does.

Speaking of this, one local audio designer is currently developing a R2R 1-bit chipless DAC-preamp design. It will be interesting to hear it once he finishes it and delivers it to the market.

It would be more interesting to see it measured by Amir. A subjective comparison could be valid when performed 'properly' (as in not sighted and level matched)

 

[mansr]

R2R 1-bit chipless DAC

Well, that's one way of saying voltage divider.

 

[zalive]

Well, that's one way of saying voltage divider.

The other guy says it can't be both R2R and 1-bit...I'll check lol.

 

[solderdude]

You need to understand technical humor to get it.

However, when you switch one resistor in an R2R configuration fast enough and are using noise shaping it could perform better than a full R2R DAC.