Serious Question: How can DAC's have a SOUND SIGNATURE if they measure as transparent? Are that many confused?

[fpitas]

I work for myself, currently I work with Sonore on product development, and production. But I am not an employee. Unless I post specifically about Sonore products (which I do not think I have done here, but I might not remember something), I am posting here as a passionate audiophile, without any agenda (promotion, etc) other than to try and learn and understand more about audio component performance. I will say, my position in the Industry does mean I am exposed to (and get to listen to) a lot of different gear, and I have worked in this industry for many years now.
For the most part I appreciate what I understand is the mission of this site, but sometimes I feel some people here go too far in terms of simplifying things in ways which I do not find to be valid. Observation is the first part of the scientific method, in our case, that would be listening. I find it a bit disheartening though, the lack of respect at this site, which is often given to those who might say things which do not agree with the chorus here ("all DACs sound the same", etc). I think if we are really pursuing science, then we should respect peoples' observations (observations are part of science), and then try and use measurements to understand why people hear what they do. The reduction of sonic performance into a relatively simple set of measurements which include only static tones is lacking IMO, and rather than just accept that this simple set of measurements is enough to describe all aspects of audio component performance, I would rather that a more open minded approach could lead to actual advancement in understanding and the development of further measurements.

I think if you're an actual design engineer, it's irresponsible to post hand-waving arguments about things we don't measure. You need to come up with something that makes sense.

 

[danadam]

Suffice it to say, life is too short to argue about what has, for me, already been addressed long ago.

A more interesting discussion I would love to see happen here might be: what measurements might actually be able to describe these differences?

Life is too short to come up with ideas for measuring something that likely doesn't exist

 

[fpitas]

Life is too short to come up with ideas for measuring something that likely doesn't exist

Something he has decided exists, so we get to figure out how to measure it. Sounds very familiar.

 

[BDWoody]

I am posting here as a passionate audiophile, without any agenda (promotion, etc) other than to try and learn and understand more about audio component performance.

Ok, taking you at your word, here's step number one.

Set up a properly controlled blind test so you can try to better support your starting premise.

Hmmm. I know that some at this site will disagree, but to my ears, there are still notable differences between DACs, despite SINAD numbers which indicate no audible difference.

I'll assume you are oversimplifying/misrepresenting for effect rather than ignorance, but if you test yourself between two DACs that have been reasonably selected for the job at hand based on the suite of measurements performed here, one of two things will result. The one I'd bet on is that you'll find those differences have disappeared. The alternate is that you will be the first ever to do it, and then we can talk more about what needs to be added.

My standard qualifiers are it has to be level matched, double blind, using normal music in a normal room or with headphones at any reasonable listening level. Basically, no extreme gain riding to elevate otherwise inaudible noise between tracks, or other tricks to game the test.

You'd think someone would have been able to do more than just double down with yet more claims, but not so far.

Are YOU open minded enough to put yourself to this simple test before telling us all we must be missing something?

 

[Mr. Widget]

I am not interested in "convincing" anyone. I am mature enough to know that would be a fools errand. I know what is right, and what is not, and I could care less if some folks are wrong. What I am interested in is moving forward, and the development of more sophisticated measurements which might attempt to describe sonic differences which are not apparent in the standard set of measurements done by Amir.

Very unlikely that that will happen. The general consensus by the objectivists is that all that needs to be known is known and any contrarian view of the subjectivists is based on ignorance. The subjectivists "know" what they have heard and it does not align with what the measurements show.

Both camps seem to be fully cemented in their knowledge that they are correct and the other side is ignorant or deaf.

Like you, I believe I have "heard" differences in devices that the measurements do not support. I have also "heard" differences that I know for a fact were not there. I think the listening part of the discussion is where we must look and requires more study. If we can prove that there are audible differences in devices that measure stunningly similarly then it would merit deeper investigation into measurement methodology.

Most of the objectivists I have discussed this general topic with are convinced that no verifiable audible differences will be found when properly tested. I can not say they are wrong, but I am very curious about the topic.

EDIT: I was writing this and working (still not retired) while @BDWoody was posting. My post is echoing his.

 

[fpitas]

If we can prove that there are audible differences in devices that measure stunningly similarly then it would merit deeper investigation into measurement methodology.

Exactly that. And I'll note that Amir has at times added measurements that made engineering sense. I'd hope a design engineer would suggest measurements, not shrug and tell us to go fish.

 

[GXAlan]

I would like to encourage discussions regarding what other measurements might be done to explain these differences objectively. … I certainly believe, that given our ability to measure extremely low level phenomena, we should be able to measure any aspect of electronic component performance which would be audible, but I do not believe we actually do so with the measurements commonly done.

Null testing is a great way to start to do these measurements with real music. As we build a library of differences between things we think are audible and backed by ABX testing, we can start to better understand what we like.

 

[sonder]

My standard qualifiers are it has to be level matched, double blind, using normal music in a normal room or with headphones at any reasonable listening level. Basically, no extreme gain riding to elevate otherwise inaudible noise between tracks, or other tricks to game the test.

You'd think someone would have been able to do more than just double down with yet more claims, but not so far.

Are YOU open minded enough to put yourself to this simple test before telling us all we must be missing something?

I'm learning, yet technical, you seem trust-able.

Broadly, are you saying that any recent-ish (good? or just any?) are broadly equivalent and indistinguishable to each other. Is there any qualification to that?

I'm sitting here with the same roon stream to 2x chromecast audio's, one running through optical to a recent topping dac, the other rca out so using the cca's internal dac chip, to the same amp and speakers, as I flick between the two inputs the difference is definitely audible to myself and everyone in the house.

So what I'm looking to establish here, is if the cheap aged dac chip in the cca is simply just that cheap and aged and not as clear as the new one. Or if there's a viable upgrade path such a noticeable difference going for another new higher priced dac (e70 velvet, a chord, or suchlike).

Or, would you be suggesting that my and our ears are wrong, or that there's some other part of the signal flow that's making the difference we hear?

What are the moving parts here.. I'm working on the assumption that once you get past a certain quality and transparency, any gains are imperceivable/ negligible.

I guess a second part to this would be, why do ESS/AKM etc have different chips, if they're all indistinguishable? Or do you need very specific high grade equipment to hear or notice a difference - or can it only be measured using equipment?

 

[Killingbeans]

or that there's some other part of the signal flow that's making the difference we hear?

Yes. Signal amplitude. There's no strict rules for the gain in DAC output stages. One can easily be more "hot" than the other, making level matching by ear highly unreliable.

 

[BDWoody]

Or, would you be suggesting that my and our ears are wrong, or that there's some other part of the signal flow that's making the difference we hear?

No, it's likely there is a real difference you are hearing in the situation you are describing. One is almost surely louder than the other, and that by itself will generally lead to the 'better' conclusion. To really test yourself, can you find a multimeter that you can use to check the voltages on the speaker cables to match a signal sent to each? Use something like a 700Hz tone at maybe -15dB, and adjust levels by reducing the higher one until it matches the lower one. Trying to do it by ear is going to be pretty much impossible, and using a phone app has way too many variables to get it within 1%. I can't remember if the 'high dynamic range' setting on the analog out on the CCA has higher noise/THD or not, but I believe it sets the output voltage to a higher level so you get a stronger signal, not necessarily a cleaner one, but I'd probably set it using the higher output to try to be as fair as possible to the other DAC. If you can then switch between them in a way that you don't know which you are listening to, you can see if these minimal controls make it any tougher.

What are the moving parts here.. I'm working on the assumption that once you get past a certain quality and transparency, any gains are imperceivable/ negligible.

This thread is a very good introduction to what might matter and what likely doesn't. A full read through the thread will help with the climb up the learning curve, at least it did for me.

Audibility thresholds of amp and DAC measurements

Introduction A recurring theme on ASR is whether or not the various measured qualities of the devices are audible. In this post, I'll present some clear and visual thresholds for when those imperfections can be considered a potentially audible concern. I will not explain the basics of amp/DAC...

www.audiosciencereview.com

I guess a second part to this would be, why do ESS/AKM etc have different chips, if they're all indistinguishable? Or do you need very specific high grade equipment to hear or notice a difference - or can it only be measured using equipment?

They likely have improved measured performance, but beyond some point, those improvements are going to be way below our ability to detect a damned thing.

We haven't referenced the Shout-o-Meter lately, and while it clearly doesn't include atmospheric attenuation (which is quite frequency/humidity/etc dependent), but it is meant to give some perspective on just how small these noise and distortion numbers are.

The Shoutometer

We've all seen the Sound Pressure Level decibel reference tables where you have "quiet room", "chainsaw", "jet engine" and so on. While that is handy for estimating increased loudness in decibles, it didn't help me when thinking about the measurement levels obtained around here when measuring...

www.audiosciencereview.com

 

[sonder]

I believe it sets the output voltage to a higher level so you get a stronger signal, not necessarily a cleaner one,

That, would make a great a deal of sense. Thank you for the thread pointer, I'll have a thorough read.
I already "felt" that moving from an AK4493SEQ dac (E30II) to a 4x more expensive AK4499EXEQ based one (E70 Velvet) would provide little or no perceivable benefit. I will guess for now that it's more about what a DAC device does to the signal after it's been converted to analogue that makes the difference, as opposed to the actual chip itself, and that if a device has a clean enough output already, and the required functionality/inputs/outputs, that the exact chip is irrelevant - discounting of course support for dsd512./pcm768 or specifics like MQA of course.

 

[BDWoody]

That, would make a great a deal of sense. Thank you for the thread pointer, I'll have a thorough read.
I already "felt" that moving from an AK4493SEQ dac (E30II) to a 4x more expensive AK4499EXEQ based one (E70 Velvet) would provide little or no perceivable benefit. I will guess for now that it's more about what a DAC device does to the signal after it's been converted to analogue that makes the difference, as opposed to the actual chip itself, and that if a device has a clean enough output already, and the required functionality/inputs/outputs, that the exact chip is irrelevant - discounting of course support for dsd512./pcm768 or specifics like MQA of course.

There are things that are important to consider which can make a huge difference in how it works for you, like the things you mention, as well as whether it has sufficient output voltage for whatever it is driving for example. Not every DAC is the right choice for every application, but generally speaking the differences in sound from one decent DAC to another is going to be dwarfed by pretty much everything else in your system other than cables.

As a general rule of thumb, if it doesn't have moving parts it is more about understanding the basics. The stuff with moving parts (transducers) is a different story...

 

[sonder]

whether it has sufficient output voltage for whatever it is driving for example.

Something else I don't know here, sigh.

DAC Output Level: 2.1 vrms
Amp X Input Sensitivity: 1.75 vrms
Amp Y Input Sensitivity: 2.6 vrms

Are these the figures I should be looking at and trying to match here in some way? Mentioning specific examples in case X or Y is incompatible. More reading required I think.

 

[Chrispy]

Something else I don't know here, sigh.

DAC Output Level: 2.1 vrms
Amp X Input Sensitivity: 1.75 vrms
Amp Y Input Sensitivity: 2.6 vrms

Are these the figures I should be looking at and trying to match here in some way? Mentioning specific examples in case X or Y is incompatible. More reading required I think.

DAC output won't drive amp Y to full power.

 

[BDWoody]

Something else I don't know here, sigh.

DAC Output Level: 2.1 vrms
Amp X Input Sensitivity: 1.75 vrms
Amp Y Input Sensitivity: 2.6 vrms

Are these the figures I should be looking at and trying to match here in some way? Mentioning specific examples in case X or Y is incompatible. More reading required I think.

The input sensitivity is the voltage required to drive the amplifier to its maximum output. The DAC in your example would be able to drive Amp X to its full output, but if you haven't attenuated the DAC output so it doesn't overload the Amp input it might clip. Amp Y will not be able to reach its full output power. Amps themselves may have different levels of gain as well, so one more thing to consider.

 

[sonder]

DAC output won't drive amp Y to full power.

a question I won't write, because it was answered by:

he DAC in your example would be able to drive Amp X to its full output, but if you haven't attenuated the DAC output so it doesn't overload the Amp input it might clip.

I feel like this entire conversation is going to reduce to using xlr/trs balanced at the right voltage level to get a perfect match?

Cracking community you have here, so many experienced members providing concise answers, thank you so much.

 

[Chrispy]

a question I won't write, because it was answered by:


I feel like this entire conversation is going to reduce to using xlr/trs balanced at the right voltage level to get a perfect match?

Cracking community you have here, so many experienced members providing concise answers, thank you so much.

No, you don't have to match them, but likely better to have too much than too little output voltage.

 

[Galliardist]

I work for myself, currently I work with Sonore on product development, and production. But I am not an employee. Unless I post specifically about Sonore products (which I do not think I have done here, but I might not remember something), I am posting here as a passionate audiophile, without any agenda (promotion, etc) other than to try and learn and understand more about audio component performance. I will say, my position in the Industry does mean I am exposed to (and get to listen to) a lot of different gear, and I have worked in this industry for many years now.
For the most part I appreciate what I understand is the mission of this site, but sometimes I feel some people here go too far in terms of simplifying things in ways which I do not find to be valid. Observation is the first part of the scientific method, in our case, that would be listening. I find it a bit disheartening though, the lack of respect at this site, which is often given to those who might say things which do not agree with the chorus here ("all DACs sound the same", etc). I think if we are really pursuing science, then we should respect peoples' observations (observations are part of science), and then try and use measurements to understand why people hear what they do. The reduction of sonic performance into a relatively simple set of measurements which include only static tones is lacking IMO, and rather than just accept that this simple set of measurements is enough to describe all aspects of audio component performance, I would rather that a more open minded approach could lead to actual advancement in understanding and the development of further measurements.

You speak with all the passion and authority of the reviewer who carefully listed all the sonic differences of the filters in the then new PS DirectStream DAC... before those filters had been written and downloaded to the player.

The point is that all of those subjective differences we read here, cannot be taken to be from the audio system (source/amp/speakers and whatever) but are of a system that consists of the hardware, source recording, and the listener. The output I get when I read a review, is the description of the reviewer's interpretation of the sound: and people we may consider reliable (most typically ourselves) certainly are not.

The proposal around competent DACs is that if we remove all the known audible differences, and run a comparison with a test subject that is blind to the system and what is under test, that the test subject will hear no difference. In order for there to be anything else to hear, then you should be able to run that test and show that a test subject hears a difference. Any other test relies not on the subject's/your hearing but their/your interpretation of the sound. It seems that our brains subconsciously process sound using past experience and other inputs like what we can see.

Observations are a part of science. There's also science involved in understanding that subjective process. First understand the nature of the observation.

 

[raif71]

DAC output won't drive amp Y to full power.

I suppose if the DAC is able to drive both amps to reasonable hearing level say 80db, that should be ok. However 80db on a single tone may be different than driving a whole range of music frequencies. That may introduce differences, yes ?

 

[Chrispy]

I suppose if the DAC is able to drive both amps to reasonable hearing level say 80db, that should be ok. However 80db on a single tone may be different than driving a whole range of music frequencies. That may introduce differences, yes ?

I can't think of why that would be a particular issue, and if not clipping the amp.....

You could use amp Y still, you just won't be able to use it fully.