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

[knownothing]

I believe some people find it difficult to figure out witch pieces of equipment is properly executed, and witch are not....

It’s all due to the presence of witches. Get the witches out of the audio chain and it’s all good.

 

[Galliardist]

This is not from speakers or headphones with recorded sound. Never mind that most of those recording were created in a studio and are totally artificial.

I guess they could take that model and use it on different types of reproduction system, though, and correlate the results with what people claim to hear.

I’d guess it would point to the speakers a lot more than it would fall for any illusion from a stereo.

If it does get used as a design tool though I’d suspect it to be used on AR/VR systems rather than boring old stereo.

 

[knownothing]

No! The article quotes the researcher:

This is not from speakers or headphones with recorded sound. Never mind that most of those recording were created in a studio and are totally artificial.

Bear with me here. The MIT model was designed to replicate the capabilities of human hearing, and the researchers claim that is difficult and that the have modeled a reasonable facimile. Audio enthusiasts listen to recorded music on their systems and strive to approximate live performance (or at least faithfully reproduce the recording engineer’s decisions in the studio). If there is spatial information in the recording that could be deciphered by a human listener, then it should be deciphered by the MIT model too. And I assume they have some way to at least partially quantify the performance of their approximation.

Take a hifi system to their lab with several different DACs, leaving all other gear the same, and see if Dumbo the model can detect differences in spatial cues from the different DACs. You should be able to replicate this test for any gear type - even cables…

kn

 

[voodooless]

Take a hifi system to their lab with several different DACs, leaving all other gear the same, and see if Dumbo the model can detect differences in spatial cues from the different DACs. You should be able to replicate this test for any gear type - even cables…

It’s totally pointless for reasons me and others have pointed out multiple times. You can imagine differently all you want, it doesn’t change the facts.

 

[Blumlein 88]

Bear with me here. The MIT model was designed to replicate the capabilities of human hearing, and the researchers claim that is difficult and that the have modeled a reasonable facimile. Audio enthusiasts listen to recorded music on their systems and strive to approximate live performance (or at least faithfully reproduce the recording engineer’s decisions in the studio). If there is spatial information in the recording that could be deciphered by a human listener, then it should be deciphered by the MIT model too. And I assume they have some way to at least partially quantify the performance of their approximation.

Take a hifi system to their lab with several different DACs, leaving all other gear the same, and see if Dumbo the model can detect differences in spatial cues from the different DACs. You should be able to replicate this test for any gear type - even cables…

kn

Their model will have a certain level of timing discrimination needed for it to work. They'll know what it is. When they find DACs are magnitudes better than that they will have no interest in wasting time testing that idea. Chances are they use DACs and ADCs in their work. So if those had timing issues it would interfere with their model and testing. They likely have no concerns about that part of it.

 

[theREALdotnet]

It is not imaginary. It is an illusion which can be demonstrated.

Indeed. A lot of people confuse these two.

 

[Blumlein 88]

Found this from one of the links interesting:

The Ventre-Parry Effect​

The Ventre-Parry Effect is a phenomenon that demonstrates how visual cues can influence the perceived distance of sounds. When individuals watch a video of somebody shouting, their perception is that the person is shouting from a farther distance than when listening to an audio recording of the same shout without the visual cue.

 

[Dimitri]

All we need is a way of measuring this:
Listen to it with your eyes closed to avoid what @Blumlein88 pointed out

 

[knownothing]

It’s totally pointless for reasons me and others have pointed out multiple times. You can imagine differently all you want, it doesn’t change the facts.

You are prejudging an outcome of an experiment that has not been run yet. We’ll agree to disagree on theory and principles until someone proves one of us wrong empirically.

kn

 

[pablolie]

I see alot of people talking about different soundstages from dacs and headphone amps. Is this similar to headphone cables, E. G. Snake oil? If it isn't why isn't it measurable?

The "stage" is included in the various measurements. If the measured response is near identical to the reference used for measurements, it means the stage resolution is there.

 

[voodooless]

You are prejudging an outcome of an experiment that has not been run yet. We’ll agree to disagree on theory and principles until someone proves one of us wrong empirically.

Not at all. You can try yourself:

What does jitter ACTUALLY sound like? (jitter test tracks)

The following website has some simulated jitter test tracks: What does jitter ACTUALLY sound like? There are seven versions of each sample. Three have periodic jitter, three have random jitter, and one is the original - untouched. In the samples we have 2µs (microseconds), 4µs, 8µs, 16µs...

www.audiosciencereview.com

Good luck, you’ll need it!

 

[Blumlein 88]

You are prejudging an outcome of an experiment that has not been run yet. We’ll agree to disagree on theory and principles until someone proves one of us wrong empirically.

kn

The best known interaural timing difference perceptible in humans is around 10 microseconds. That will be around 1200-1500 hz and gets worse at higher frequencies. From memory I think it almost disappears above 2500 hz. It gets slightly worse at lower frequencies.

Here is an abstract of fairly recent tests of that. It isn't out of line with similar testing done decades ago.

Smallest perceivable interaural time differences - PubMed

It is well-established that the smallest discrimination thresholds for interaural time differences (ITDs) are near 10 μs for normal hearing listeners. However, little is known about the hearing and training status of the test subjects from past studies. Previous studies also did not explicitly...

pubmed.ncbi.nlm.nih.gov

10 microseconds is a million times longer than 10 picoseconds.

The Neural network model you linked references a few articles detailing the same thing.

 

[voodooless]

Also don’t see how this would somehow translate into bigger or smaller soundstage. Also note that for this to work, timing differences of signals would have to differ per channel, and even per source. It just doesn’t make sense.

 

[FrantzM]

Surprised this is still going on...

Out

Peace,

 

[voodooless]

Surprised this is still going on...

What do you mean? We’re only at page 4

 

[mjgraves]

In browsing this thread there's a lot of mention of the timing of signals. What is really of concern is differential (relative) timing. Would the device (DAC or whatever) significantly alter timing of the two channels substantially differently? I think not.

 

[allmanfan]

Perhaps not snake oil, but certainly not verified in a proper, scientific, blind, ABX listening test. So yes, I'd say it's similar to the imaginary differences people hear with cables.

Some headphone amps have optional crossfeed which blends the left & right channels. That WILL affect soundstage.

If one listens to different HP's and cannot distinguish different soundstages in terms of width and depth then I truly wonder about that persons ability to discern because it is blatant and obvious....DAC's and amps also will have more or less bass, different soundstages etc....it is not nuanced like cables it is very apparent

 

[theREALdotnet]

The transfer fidelity of any piece of electronics can be verified with a null test, comparing the before and after waveforms. Any differences, regardless of whether they were caused by linear or non-linear distortion, noise, jitter, any kind of phase shenanigans, would show up there. If a DAC were to mess with phase or timing at certain frequencies, then this would very clearly show up in a null test, since the waveforms would be quite different.

 

[DMill]

DAC's and amps also will have more or less bass, different soundstages etc....it is not nuanced like cables it is very apparent

DACs and Amps are not intended to be EQ solutions. I suppose we could design an amp that accentuates bass, but why? Also there is no nuance to cables that aren‘t defective in some way. Soundstage, is happening in the recording.

 

[knownothing]

Also don’t see how this would somehow translate into bigger or smaller soundstage. Also note that for this to work, timing differences of signals would have to differ per channel, and even per source. It just doesn’t make sense.

I am not saying and I’m not sure jitter is the potential source of small differences in time, loudness and frequency domains in different DACs that could lead a measurement or a listener to perceive different spatial cues in the listening environment. ESS DAC chips have a particular sound, but all electronics utilizing a particular model DAC chip do not sound identical. There are decisions designers make in implementing circuits, choosing power supplies, designing shielding, etc. that can affect the resulting sound. It’s 0’s and 1’s, until it isn’t.