Master Thread: Are Measurements Everything or Nothing?

[DavidM1]

Being somewhat out of touch these days, can someone summarize validated audibilty for the effects under discussion?

There are explanations elsewhere here and they are always interesting to read.

Tangent:
I think of hearing like vision. As a kid I had better than 20/20 but these days I wear glasses for driving or going to the movies.

Audio test equipment is like a microscope that reveals a level of detail invisible to anyone’s naked eye. It would be self-deluding to think mere eyesight, no matter how experienced, could compete.

I’m happy to trust the numbers on audibility and I’m grateful to this site for the amount of objective testing it publishes.

 

[ahofer]

What really makes zero sense is folks spending tens to hundreds of thousands of dollars on speakers but then being unwilling to do most anything with the room. Bare walls, hardwood floors, walls of windows with a wonderful view, but with a sound almost identical to their shower stall.

Hey, I sound great in my shower.

 

[satow]

@ satow: Why are you using a Shiit Freya preamp when you have a Gustard DAC with variable volume ? You are so injecting distortion and noise into the amp, so why having a nearly perfect Purifi amp for using it that way ? Glad to hear your point, thanks in advance

I use the Schiit Freya+ F preamp because it adds the spatial depth, dimensionality, and harmonic realism that a DAC’s digital volume control cannot provide. The Gustard R26 DAC's variable output is excellent, but running it direct sounds flatter and less holographic in my system. The Freya+ in tube mode introduces the exact kind of euphonic shaping that helps my Magnepan LRS disappear and form a true 3D image, while the Apollon Purifi 9040 amp produces ultra‑low distortion which ensures that this character is preserved without adding noise. It’s a deliberate choice for imaging and musicality, not a technical mistake. Also next month, hopefully I can add a Gustard U18

 

[Ropeburn]

I use the Schiit Freya+ F preamp because it adds the spatial depth, dimensionality, and harmonic realism that a DAC’s digital volume control cannot provide. The Gustard’s variable output is excellent, but running it direct sounds flatter and less holographic in my system. The Freya+ in tube mode introduces the exact kind of euphonic shaping that helps my Magnepan LRS disappear and form a true 3D image, while the Purifi’s ultra‑low distortion ensures that this character is preserved without adding noise. It’s a deliberate choice for imaging and musicality, not a technical mistake.

All of which you describe is in the music, and brought out by speakers in a room, to varying degree. A simple electronic device doing nothing but amplifying or attenuating the line signal can't change anything about these sonic properties.

3D image, holography, dimensionality, these are emergent properties of sound design and mixing, not of a simple preamp.

 

[Ropeburn]

Oh I forgot: "euphonic shaping"... yeah, that's distortion. Another thing you get from the music but don't want from your playback system.

You like distortion? I'll give you distortion!

6581 Terminator X.flac | Powered by Box

app.box.com

"Euphonic shaping" indeed! The distortion on that patch is envelope controlled (=shaped), that's why it's so gnarly and still "round". Spatiality comes from simple short room effect.

That's the kind of distortion you actually want to hear if you're into that kinda stuff. Not a deliberately flawed playback system.

 

[satow]

Oh I forgot: "euphonic shaping"... yeah, that's distortion. Another thing you get from the music but don't want from your playback system.

You like distortion? I'll give you distortion!

6581 Terminator X.flac | Powered by Box

app.box.com

"Euphonic shaping" indeed! The distortion on that patch is envelope controlled (=shaped), that's why it's so gnarly and still "round". Spatiality comes from simple short room effect.

That's the kind of distortion you actually want to hear if you're into that kinda stuff. Not a deliberately flawed playback system.

LOL! I get what you’re saying, but the idealized “perfectly transparent gain block” doesn’t describe how real analog stages behave. Spatial cues come from the recording, but different circuits preserve or collapse those cues to different degrees because of phase behavior, noise modulation, harmonic structure, and load interaction.

Nobody’s claiming a preamp creates holography. The point is that some designs maintain microdetail and depth better than others.

And yes—your FLAC example shows exactly why the type and envelope of distortion matters. That same principle is why different gain stages can sound different even when they measure clean.

 

[RexrothPigeon]

LOL! I get what you’re saying, but the idealized “perfectly transparent gain block” doesn’t describe how real analog stages behave. Spatial cues come from the recording, but different circuits preserve or collapse those cues to different degrees because of phase behavior, noise modulation, harmonic structure, and load interaction.

Nobody’s claiming a preamp creates holography. The point is that some designs maintain microdetail and depth better than others.

And yes—your FLAC example shows exactly why the type and envelope of distortion matters. That same principle is why different gain stages can sound different even when they measure clean.

Nice AI response, I'm sorry to say it's not true

 

[satow]

Nice AI response, I'm sorry to say it's not true

Your comment is fair, but the explanation isn’t AI‑generated — it’s just basic analog behavior. A DAC’s digital volume control is clean, but running direct often flattens depth because the output stage isn’t optimized to drive long interconnects or complex loads. A good preamp doesn’t “add holography,” it simply preserves micro‑dynamics, phase integrity, and spatial cues that can get lost when the DAC is used as the sole gain stage. The Schiit Freya+ F preamp happens to maintain those cues better in my system, which is why the Apollon Purifi 1ET9040BA DM Dual Mono Stereo amp actually reveals more depth and realism with it in the chain, not less. That's why my system sounds fantastic, you can hear people walking around on the stage, not just left and right, but forward and back. Instruments are easily placed in the generated soundstage. Tube rolling is a good example of why the “all clean gain stages sound identical” argument doesn’t hold up in real systems. Different tubes measure similarly at 1 kHz, yet anyone who has rolled 6SN7s knows they vary in imaging, depth, tone density, and spatial realism because of differences in microphonics, noise behavior, transconductance, and harmonic structure. The same principle applies to analog output stages and preamps — small differences in how they handle phase, noise, and load interaction can preserve or collapse spatial cues, even when the bench measurements look nearly identical. You can ask any musician who uses an amp. They tube roll for a reason. Just like in my preamp. Even if you are using high quality matching tubes or stock tubes, they will all sound different. My trick for the Schitt Freya +F preamp is to use certain tubes. PSVANE Summit 6SN7 tubes for the gain and NOS GE 6SN7‑GTB for the Cathode followers.

 

[Ropeburn]

1) What line input stage is a complex load? Certainly not an opamp buffered one. Their whole point is providing a high impedance "low complexity" load. Resistive.

2) Any preamp can't possibly "preserve" any information better than no preamp. DAC -> preamp -> poweramp simply cannot be more true to the original signal than DAC -> poweramp. That's physically and logically impossible. You know that, at the very least subconsciously: you literally wrote the preamp adds all these things. Hehe.

You already know the truth, you just don't want to, it seems.

 

[satow]

Regarding section1 "complex load”:
A buffered op‑amp input is high‑impedance, but it is not a purely resistive, zero‑interaction load. The DAC still drives: finite input impedance, cable capacitance and inductance, dielectric absorption, the op‑amp’s input bias network, and any RF filtering at the input. Those elements form a real transfer function that affects HF phase, settling behavior, and noise modulation. That’s why different line stages and cable/interface combinations can measure identically at 1 kHz yet differ in wideband behavior and time‑domain performance.

and regarding number 2 “no preamp is always more accurate”:
“DAC → amp is always truer” is only valid in an ideal system with zero output impedance, infinite input impedance, zero cable reactance, and a DAC output stage designed purely as a line driver. In practice, many DACs have modest output stages optimized for short, easy to drive loads, not for driving longer interconnects or more complex input networks. A dedicated preamp with lower output impedance, higher current capability, better common‑mode and RF rejection, and cleaner wideband behavior can reduce interface‑induced errors (HF phase shift, noise modulation, level‑dependent distortion). It doesn’t “add information”; it reduces the degradation that occurs when the DAC is forced to do both precision conversion and heavy line‑driving at the same time. That is why my Gustard R26 DAC replaced my SMSL SU-10 DAC. It sounded better.

I would purchase a Rockna Wavedream DAC, but I ain't rich like you guys.

 

[Waxx]

some people like added harmonic distortion, that is a fact. It's distortion, but if it pleases the listener, why would you be against it. It's a deviation from high fidelity in the strickt sense of the word, but if it brings more joy to listen to music, it's a good choice. And Satow is not the only one who does that, many do. It's a subjective preference, not an objective quality and that is not bad as long as you know it's subjective and an distortion.

 

[antcollinet]

but different circuits preserve or collapse those cues to different degrees because of phase behavior, noise modulation, harmonic structure, and load interaction.

If you have something that does that you should throw it away (don't sell it, the buyer doesn't deserve that put on them) and replace it with one of the other 99% of competent devices that don't do it.

Even half way decent devices don't have noise or distortion at anywhere near the levels needed to cause audible effects such as those you describe.

Further - "complex loads" (Other than speakers/'phones) don't exist - at least not sufficiently to have an audible impact. (capacitance/inductance of cables and input stages is too small to create audible artefacts)

 

[RandomEar]

This is turning into another "user claims audible XY which doesn't exist"-discussion. That would be fine, if it weren't very off-topic because this is a buying experience thread. Could you maybe continue this in a separate one?

 

[satow]

If you have something that does that you should throw it away (don't sell it, the buyer doesn't deserve that put on them) and replace it with one of the other 99% of competent devices that don't do it.

Even half way decent devices don't have noise or distortion at anywhere near the levels needed to cause audible effects such as those you describe.

Further - "complex loads" (Other than speakers/'phones) don't exist - at least not sufficiently to have an audible impact. (capacitance/inductance of cables and input stages is too small to create audible artefacts)

Real audio circuits don’t behave like ideal textbook blocks, and small differences in wideband phase shift, noise modulation, clock behavior, and output‑stage loading can alter low‑level spatial cues even when two devices measure the same on a 1 kHz SINAD test. A simple example is output‑stage settling time: two DACs can have identical distortion numbers, yet if one settles more slowly or less linearly during fast transients, it will smear micro‑timing information between channels. That distortion is tiny, but because it’s correlated with the music rather than random noise, it can change imaging and depth without showing up in the usual steady‑state measurements people cite.

 

[Ropeburn]

Oh!

This moving posts here makes a lot of sense. Thumbs up!

I do try not to derail topics but sometimes I can't help it. Please forgive your friendly neighbourhood Doge

 

[antcollinet]

Real audio circuits don’t behave like ideal textbook blocks, and small differences in wideband phase shift, noise modulation, clock behavior, and output‑stage loading can alter low‑level spatial cues even when two devices measure the same on a 1 kHz SINAD test. A simple example is output‑stage settling time: two DACs can have identical distortion numbers, yet if one settles more slowly or less linearly during fast transients, it will smear micro‑timing information between channels. That distortion is tiny, but because it’s correlated with the music rather than random noise, it can change imaging and depth without showing up in the usual steady‑state measurements people cite.

I'm sure you can point to measurements and/or well controlled blind tests as evidence for this nonsense.

Audible "settling time" would have massive effect on measured distortion. But it isn't a thing, as demonstrated by the fact that DACS can produce perfect sine waves many times higher frequency than the highest frequency any human can hear.

 

[Ropeburn]

I'm sure you can point to measurements and/or well controlled blind tests as evidence for this nonsense.

Tee hee hee. I'm sure he/she/it can't.

 

[satow]

I'm sure you can point to measurements and/or well controlled blind tests as evidence for this nonsense.

Audible "settling time" would have massive effect on measured distortion. But it isn't a thing, as demonstrated by the fact that DACS can produce perfect sine waves many times higher frequency than the highest frequency any human can hear.

I’m just saying that steady‑state sine tests don’t show everything. Two DACs can match on a 1 kHz SINAD number yet behave a little differently with real music because of things like phase behavior, noise modulation, or how the output stage settles after fast transients. Those effects are tiny, but they’re signal‑dependent, so they don’t show up in simple sine‑wave measurements even though they can influence imaging. It’s just normal time‑domain behavior, not anything exotic.

 

[satow]

Try this in a simple experiment with two DACs that measure the same on a 1 kHz SINAD test but use different reconstruction filters.

First, pick DAC A with a slow‑rolloff linear‑phase filter, and DAC B with a minimum‑phase filter - verify that both show essentially identical THD+N and SINAD at 1 kHz.

Next, feed them an impulse or fast step and capture the output on a scope or use DAW software:

You will find that DAC A will show pre‑ringing and a longer impulse response, while DAC B will have no pre‑ringing and will settle faster after the transient.

Then run a multitone or wideband test and look at the time‑domain and FFT plots—you’ll see differences in how each DAC handles transients and energy distribution even though the 1 kHz SINAD number is the same.

Now listen to some well known music in a level‑matched, quick‑switch setup: those small, signal‑correlated differences in transient behavior can show up as subtle changes in audio imaging and clarity, despite both devices looking “identical” on the basic sine‑wave spec sheet.

 

[IPunchCholla]

Try this in a simple experiment with two DACs that measure the same on a 1 kHz SINAD test but use different reconstruction filters.

First, pick DAC A with a slow‑rolloff linear‑phase filter, and DAC B with a minimum‑phase filter - verify that both show essentially identical THD+N and SINAD at 1 kHz.

Next, feed them an impulse or fast step and capture the output on a scope or use DAW software:

You will find that DAC A will show pre‑ringing and a longer impulse response, while DAC B will have no pre‑ringing and will settle faster after the transient.

Then run a multitone or wideband test and look at the time‑domain and FFT plots—you’ll see differences in how each DAC handles transients and energy distribution even though the 1 kHz SINAD number is the same.

Now listen to some well known music in a level‑matched, quick‑switch setup: those small, signal‑correlated differences in transient behavior can show up as subtle changes in audio imaging and clarity, despite both devices looking “identical” on the basic sine‑wave spec sheet.

If this is a test you’ve run, can you share the data? If it is not, why are you claiming you know what it will show?