Can you hear a difference between DAC's?

[upstateguy]

Hi Guys
I've posted this a couple of times over the years, but check this out.
I have an older April music Stello DA100 Dac. On the front it has an upsample to 24/192 switch.

When you throw the switch it changes the sound. We had some lively discussions about which position sounded better. The feature was also discussed in the manual and April Music suggested you choose the position that sounded best for the music you were listening to. For a long time I thought the 24/192 upsample reduced the bass, then I used the Audio_DiffMaker to see what the difference was. I was wrong, it didn't reduce the bass, what was left after nulling was 'treble'. So the 192 upsample increased the treble which made it seem like the bass was reduced. Wait, what?

So here's the thing. If the circuitry can change the sound of a DAC, how is it possible that all DACs, with all kinds of circuitry, could sound the same, when this DAC alone can sound 2 different ways?

These days I'm using a RME ADI-2 with the AKM chip, 2 Genlec 8010As and a 7040 sub for a near field desktop, music, movies, everything. I don't use headphones much anymore but when I do, I like my original T-1s with the speaker system on so I can feel the bass.

There's more. At one of the meets a few of us did some double blind tests between that Stello DA 100 and a Benchmark Dac. We couldn't tell the difference.

Hey castleofargh how are you, long time no see.

 

[Baten]

Hi Guys
I've posted this a couple of times over the years, but check this out.
I have an older April music Stello da100 Dac. On the front it has an upsample to 24/192 switch.


When you throw the switch it changes the sound. We had some lively discussions about which position sounded better. The feature was also discussed in the manual and April Music suggested you choose the position that sounded best for the music you were listening to. For a long time I thought the 24/192 upsample reduced the bass, then I used the Audio_DiffMaker to see what the difference was. I was wrong, it didn't reduce the bass, what was left after nulling was 'treble'. So the 192 upsample increased the treble which made it seem like the bass was reduced. Wait, what?

So here's the thing. If the circuitry can change the sound of a DAC, how is it possible that all DACs, with all kinds of circuitry, could sound the same, when this DAC alone can sound 2 different ways?

Good on you for nulling but that upsampling SRC chip can be programmed in such a way that would explain the difference. So it is more DAC+SRC chip in this case, no.

 

[tlite]

Sure, what music genre you like?

Anything but pop country and smooth jazz

 

[Ferdinando1968]

It's simple, if the measurements are something like voltage, frequency, some specific type of distortion or something as silly as time, then it's unreliable, as none of those things are real music.

I would like to ask your opinion.
The measurements you are talking about can be performed in a very precise and reproducible way,
therefore they can be considered 'scientifically' and technically valid.

Personally I think that what we can perceive with our senses (listen) can be highlighted in the measurements; in other words, I don't think we can perceive anything that the instruments cannot highlight.

And therefore why shouldn't it be possible to 'classify' a certain quality or sound performance through a measurement, at least in the field of electronic equipment such as DACs
(never mind the transducers, because the matter is more complex in that field).

What or what would be the reasons?

Are they perhaps related to the fact that the above measurements are related to specific aspects of the electrical signal while what we hear is a complex or a sum of aspects?

Also: have I sometimes read something about 'synergies' between certain electronics and transducers?

In your opinion is there anything technically valid in all of this?

Shouldn't these synergies be identifiable by electrical measurements?

Thank you.

 

[NTTY]

I can hardly hear a difference between two DACs, I suspect the quality of the whisky in my glass to have much more influence on my listening experience (even before drinking it) than two DACs.

But it’s easy to hear differences (for me and whisky agnostic) between two speakers, even the same speakers in two different rooms.

It’s also very easy to hear differences between two masters which is where I’m spending my time (to find them) and money now.

I owned a number of popular DACs (RME, Gustard, …) and I got into the hobby of measuring my devices to see how far in the past I needed to go to hear a difference. Recently I was able to tell the difference between an old and obsolete Philips TDA1541A with its 4x oversampling filter, the SAA7220 and a much more recent BB PCM1792 with the advanced AL24 processing of Denon. But it was on a tiny detail, piano playing a little more on the left side on one of my preferred high quality master recording.

As I’ve seen other members mentioning it before, it is important to match levels between two devices under test, and it’s not always easy, as it should be below 0.2dB. Else the device playing the loudest will be the preferred, revealing more hidden details, better soundstage, more bass, etc…, without a surprise since that’s what happen when you pump up the volume.

In my above test experience, initially with several headsets, I had to use Foobar ABX plugin to identify a difference between the two DACs, else I failed. And I could certainly not decide which one I preferred.

When on my standard system, it was not only easier to identify the two DACs but also I could tell which one I preferred. Indeed one of them played with narrower soundstage and that was the old Philips. Other than that, I could not tell any differences, which amazed me.

When measuring the Philips based DAC, I could only see somewhat poor crosstalk (94dB @20Hz up to 74dB at 20kHz) and channel mismatch of 0.12dB to correlate with my listening experience. Rest of the measurements were of no concern, really, beautiful multitone results, low jitter, decent SNR and DR, low distortion, no IMD of concern, etc… Oh maybe the filter a bit too slow and of poor out of band attenuation (-50dB only). It could also be a reason for the impact on the soundstage, I admit I lack experience on this one.

Regards

 

[NTTY]

I would like to ask your opinion.
The measurements you are talking about can be performed in a very precise and reproducible way,
therefore they can be considered 'scientifically' and technically valid.

Personally I think that what we can perceive with our senses (listen) can be highlighted in the measurements; in other words, I don't think we can perceive anything that the instruments cannot highlight.

And therefore why shouldn't it be possible to 'classify' a certain quality or sound performance through a measurement, at least in the field of electronic equipment such as DACs
(never mind the transducers, because the matter is more complex in that field).

What or what would be the reasons?

Are they perhaps related to the fact that the above measurements are related to specific aspects of the electrical signal while what we hear is a complex or a sum of aspects?

Also: have I sometimes read something about 'synergies' between certain electronics and transducers?

In your opinion is there anything technically valid in all of this?

Shouldn't these synergies be identifiable by electrical measurements?

Thank you.

Castleofargh’s message was sarcastic and made me lol.

Yes, obviously we can measure much more that what we can hear. Today’s DAC chips decently implemented will be perfect to listen to music.

Yes there are some sort of synergies typically between amplifier and speakers as you suggest. That happens when one connects high output impedance amplifier (ie low Dampening Factor) to speakers going low in ohms and with high phase variation making then complex to drive in some frequencies with negative impact on the bandwidth’s flatness. That’s one more reason to trust measurements as they prevent such expensive mistakes to consumers.

​

 

[OnlySoMany]

'Also, a $100,000 dac will not sound like a $20,000 dac'.


'Hey Bob I just spent $100k on a new dac'.
'Does it sound better than your $20k one?'
'No mate, it's just different'.

Yeah, pretty much

Some would say blondes are better than brunettes. Some would say brunettes are better than blondes. Some would say redheads are better all day.

Just because something technically measures better doesn't ALWAYS mean that it sounds better, it's all down to preference. What I was saying is most DACS in the same price point generally sound pretty similar until you go up to a different technology or a more advanced version of that technology.

 

[Ryokan]

Yeah, pretty much

Some would say blondes are better than brunettes. Some would say brunettes are better than blondes. Some would say redheads are better all day.

Just because something technically measures better doesn't ALWAYS mean that it sounds better, it's all down to preference. What I was saying is most DACS in the same price point generally sound pretty similar until you go up to a different technology or a more advanced version of that technology.

Yeah I was just teasing you

What do you call a red head between two blondes? The translator.

 

[OnlySoMany]

Yeah I was just teasing you

What do you call a red head between two blondes? The translator.

Good one

 

[castleofargh]

ersonally I think that what we can perceive with our senses (listen) can be highlighted in the measurements; in other words, I don't think we can perceive anything that the instruments cannot highlight.

That's a certainty. If we're able to get it on a record, it literally means we could quantify it with some tools.
But!

And therefore why shouldn't it be possible to 'classify' a certain quality or sound performance through a measurement, at least in the field of electronic equipment such as DACs
(never mind the transducers, because the matter is more complex in that field).

What or what would be the reasons?

Are they perhaps related to the fact that the above measurements are related to specific aspects of the electrical signal while what we hear is a complex or a sum of aspects?

The problem is that given the right tools, we're always going to measure differences between 2 DACs. Now we have to find a way to tell what's audible and what is not. We can only do that if we have enough data on listening tests under those fairly specific circumstances. Good luck.
When you decide to involve humans in measurement classification, ranking and how they feel about certain sonic qualities and variables, be prepared to multiply the variables, their interactions and philosophical problems several times over. For example, even harmonics, they sound more natural than odd harmonics to people. A crying baby is more annoying than white noise of the same amplitude(if your DAC has crying baby noises in it on all tracks, call your local occult expert!).
Each variable will need a well accepted hearing threshold, not in an anechoic chamber, but in the average house, with average playback gear and hearing skills.
Then we need some scale that follows how objectionable certain variables or sound qualities get as their amount increases. You must create from scratch an equivalence between variables at a subjective level, and then identify them in measurements to help ranking, classification, anything you wish out of that. It's hell, but with much longer work hours.

If we only cared about objective ranking, we'd still have to invent matching scales to decide if a frequency response flatter by 0.02dB is worth as much as an extra 1kHz of ultrasounds for example. Or if -50dB hiss is as bad as -50dB THD. Rapidly, we'll have to fall back on some human determination of what's better than something else, IMO.

Also: have I sometimes read something about 'synergies' between certain electronics and transducers?

In your opinion is there anything technically valid in all of this?

Shouldn't these synergies be identifiable by electrical measurements?

They should, but again, when people judge good sound casually with subjective preference, who knows in advance what they'll declare better? As mentioned, the most significant "synergy" affair is usually between amp and transducer(because transducers have by far the widest ranges of sensi and impedance), with impacts often involving FR among other things. If it turns out to create a significant FR change for a certain combo compared to another one, as the better FR for a given listener is strongly based on his own HRTF and the FR of the headphone, a certain pairing could be better for one person with some luck and worst for most other listeners. And that even at an objective level for purely acoustic reasons.
Then comes personal preference because of what we usually use as a headphone...

Some results are just worse for everybody, and it's easy to say, "don't do it!" if we know of that particular combo. But in general, we'd require a bunch of specs that are rarely given, and someone to collect all that data for all the possible pairings that exist and will exist.
It's easy to get why nobody ventured into doing it, even if just for 2 or 3 objective variables as a basic rule of thumb helper.
Luckily, most pairings turn out to just work fine because engineers tend to follow standards (until someone from marketing convinces them to push something to 11).

For DACs, A USB that's actually near 5V if it's usb powered, and an amplifier that's fine getting however many volts the DAC outputs, and we should be good. If it was properly designed, that should take care of most situations and have nominal operation.
If your DAC outputs 4V full scale, maybe don't plug it into a portable amp designed for DAPs with outputs below 1V. Maybe it will work fine anyway and not distort, but there are probably better pairings.

 

[upstateguy]

The problem is that given the right tools, we're always going to measure differences between 2 DACs. Now we have to find a way to tell what's audible and what is not. We can only do that if we have enough data on listening tests under those fairly specific circumstances. Good luck.

Hey CofA, I'm rusty on this, but if there are measurable differences between 2 DACs, what happened to Bit Perfect?

 

[upstateguy]

Good on you for nulling but that upsampling SRC chip can be programmed in such a way that would explain the difference. So it is more DAC+SRC chip in this case, no.

Hey Baten, could you go into a little more detail. Don't most modern DACs upsample to some extent? If they do, it seems that no two DACs would sound the same unless they used the exact same chips. And like I asked Castle, years ago Bit Perfect was a big thing. If all these DACs are using differently programmed chips, what happened to Bit Perfect?

 

[REXNFX]

Anything but pop country and smooth jazz

Same piece, both on streaming services:

 

[BS5711]

I have never tried any high end DACs but have tried the usual lower to mid range stiff, Schiit Modi, Modi Multibit, Mojo, Gryphon, Diablo etc all into the same several amplifiers and I can't hear any difference that I could not hand on heart say was more likely to be due entirely expectation bias.

The only DAC that I genuinely heard a difference with was an Ali Express special PCM56 based R2R DAC that I understand was based on an old early CD era chip. That just sounded soft, mellow and frankly borderline mushy, it was more wrong than it was different.

 

[Baten]

Hey Baten, could you go into a little more detail. Don't most modern DACs upsample to some extent? If they do, it seems that no two DACs would sound the same unless they used the exact same chips. And like I asked Castle, years ago Bit Perfect was a big thing. If all these DACs are using differently programmed chips, what happened to Bit Perfect?

Problem with using another DSP chip for resampling is that it could have internal headroom, so turning on/off could change level anywhere between 0,5-3dB.
Same for ESS chip DACs allowed to turn 'off' the oversampling/OSF filter, which causes the volume to get noticeably louder.

So not talking about Bit perfect, talking about unexpected and possibly slight volume differences making it very hard to check for 'sound difference'.

Also with regards to the high frequency difference in nulling could also be explained by a terrible OS filter. I guess. Which you can also switch to in modern DACs, yes. Switcher from Fast filter to Super Slow variant is for most people audible unless you don't hear reliably up to 15/16kHz.