What set of measurements can equate two DACs?

[musicapristina]

What set of measurements can be performed on 2 DACs such that if the two sets of measurements were identical, we could accurately predict that nobody could tell them apart in an ABX test?

 

[Bullwinkle J Moose]

Ears

If ears cannot tell them apart in an ABX test, then the measurements would be irrelevant to tell you how they should or would sound different in an ABX test

Try Price?

If they both sound the same but one is 4 times the cost, then that really sounds horrible, doesn't it?

 

[musicapristina]

Clever, but I’m looking for a serious collection of measurements.

 

[617]

Clever, but I’m looking for a serious collection of measurements.

The majority of DACs reviewed on ASR have noise and distortion components which are at or beyond the threshold of hearing. The few which have noise and distortion levels high enough that they wouldn't be masked by content would be difficult to compare by measurements. ASR doesn't really characterize the sound of DACs, it simply validates their transparency.

To paraphrase Tolstoy, all good DACs are alike; all bad DACs are bad in their own way.

 

[musicapristina]

This is where I want to get to, though:
If we have 2 "validated transparent" DACs, one selling for $299 and one for $29,999, is it fair to make the claim, based on science, that the $29,999 is over-priced snake oil that sounds the same as the one for $299?

Is it true that given 2 DACs have equal Noise and Distortion measurements, we can predict that they are indistinguishable in an ABX test?

If YES, then all we need to do are measure these two factors and we can "score" DACs for Accuracy. Divide the score by price for Value.

If NO, then we're applying science in a way that is interesting, but inaccurate, unfair and maybe dangerous.

--

This seems very important to me, and I'm not seeing this concern from other people. I asked this question in a much more provocative way the other day and got shut down for it.

 

[Purité Audio]

I have compared fine measuring fairly expensive dacs to inexpensive fine measuring designs and I couldnt hear any difference, latterly the Mola-Mola Tambaqui and Soncoz SGD1.
The measurements suggest no difference and I couldn’t detect any, the more interesting question is how much distortion to you have to add before it is becomes audible.
Keith

 

[musicapristina]

@Purité Audio Thanks.

I'd love to move from your experiences (and my experiences) to something scientific.

I've been accused of searching for magic and trolling, but I'm really just looking to confirm the validity of measurements, or file them away in the category of "useful" but not the whole story.

Claims are being made here, I would think more people would want those tested / verified. Without that, how are we (the pro-measurement) better than the subjectivist (pro-measuring-with-your-ears) crowd.

 

[BDWoody]

This is where I want to get to, though:
If we have 2 "validated transparent" DACs, one selling for $299 and one for $29,999, is it fair to make the claim, based on science, that the $29,999 is over-priced snake oil that sounds the same as the one for $299?

Not necessarily snake-oil, it just won't sound different/better.

Is it true that given 2 DACs have equal Noise and Distortion measurements, we can predict that they are indistinguishable in an ABX test?

They don't have to be equal, they just have to be beyond the threshold of transparency. Beyond that, it's not about SQ any longer.

Strict threshold might be anything over (under?) -115dB sinad. Realistic threshold is likely closer to -90dB or less with normal music in a normal room.

Beyond some point, the differences in the signals are far below what could be detected with ears.

 

[Quomz]

This seems to cover audible limits in DAC measurements. Might not be a cookie cutter answer to your question though.
https://www.audiosciencereview.com/...-thresholds-of-amp-and-dac-measurements.5734/

 

[Katji]

"Implementation" - like the filters implementation?

Dual DAC chips and channels, more thorough noise isolation...? If so, then they get seriously overpriced with that. Like $500-1000 to $5000.

 

[preload]

If 2 DACs have chassis mounted RCA jacks and thick aluminum face plates, they probably sound the same.

 

[AdamG]

In addition to just pure DAC performance. We have to account for varying levels of functionality, features and attributes. A simple pure DAC against another simple pure DAC is easy comparison. But then some are also Streamers, Network Servers, Equalizers ect…Then it’s not Apples to Apples. Only the buyer can put a value on the additional functionality. Imho.

 

[musicapristina]

@BDWoody So now my question becomes:
Is it true that given 2 DACs have Noise and Distortion measurements below -115 dB SINAD, we can predict that they are indistinguishable in an ABX test?

I understand that on this metric (SINAD), they *should* be indistinguishable to human ears.

And if this is the only measurement that matters, they will be indistinguishable to human ears. Provable through ABX testing.

But are there other measurements that we're missing? We need a falsifiable hypothesis to conduct science, right?

The hypothesis then is: Given 2 DACs have Noise and Distortion measurements below -115 dB SINAD, we can predict that they are indistinguishable in an ABX test.

And if someone can create a DAC with SINAD < -115 dB that can be reliably distinguished by ABX testing, then we need new measurements.

 

[Purité Audio]

The only difference ‘may’ be the choice of reconstruction filter, try a level matched, unsighted comparison for yourself it will be revelatory.
Keith

 

[musicapristina]

@Quomz Thanks for the link. I saw it yesterday, and see now that I can extract the "measures" I'm looking for from it. Is this exhaustive?

 

[musicapristina]

@Purité Audio I've not had the privilege of hearing enough DACs side by side and knowing their measurements.

 

[Purité Audio]

You don’t need to know their measurements to compare, the mind is hugely powerful when you know which you are listening to, bias takes over, hence the need for unsighted and level matched comparisons.
In my opinion digital is done choose on function and aesthetics, and if you want. ‘pride of ownership’ product there is nothing whatsoever wrong with that but don’t allow a salesman to tell you that it is going to sound any better.
Keith

 

[BDWoody]

And if someone can create a DAC with SINAD < -115 dB that can be reliably distinguished by ABX testing, then we need new measurements.

I'd agree with that, with the caveat that they aren't gaming the system with pathological conditions or tortured test tones, in a normal room with normal music with filters matched, levels matched, and with a properly done double blind environment.

All that's needed is for someone to do it. There has been a lot of money offered, but so far hasn't happened.

Given the lack of contrary evidence to date, it seems to be a relatively safe operating premise.

 

[Jimbob54]

@BDWoody So now my question becomes:
Is it true that given 2 DACs have Noise and Distortion measurements below -115 dB SINAD, we can predict that they are indistinguishable in an ABX test?

I understand that on this metric (SINAD), they *should* be indistinguishable to human ears.

And if this is the only measurement that matters, they will be indistinguishable to human ears. Provable through ABX testing.

But are there other measurements that we're missing? We need a falsifiable hypothesis to conduct science, right?

The hypothesis then is: Given 2 DACs have Noise and Distortion measurements below -115 dB SINAD, we can predict that they are indistinguishable in an ABX test.

And if someone can create a DAC with SINAD < -115 dB that can be reliably distinguished by ABX testing, then we need new measurements.

Dont forget the filters. Small but possibly audible differences in FR due to filters. But assuming filters in the 2 dacs have similar properties, should still be indistinguishable.

 

[ebslo]

So now my question becomes:
Is it true that given 2 DACs have Noise and Distortion measurements below -115 dB SINAD, we can predict that they are indistinguishable in an ABX test?

And if someone can create a DAC with SINAD < -115 dB that can be reliably distinguished by ABX testing, then we need new measurements.

Consider a device constructed from a low-noise high-Q 1kHz sinusoidal oscillator. The device takes a PCM input which it rectifies and LP filters to well below 1kHz, which is then used to set the amplitude of the generated 1kHz tone. This device would score very well on the DAC SINAD test despite the fact that it's not a DAC at all! You need at least a frequency response test and probably also the IM distortion test to show that a device is even a DAC.

Edit: Forgot the linearity test, it's very important too.