• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

How replicate the sound of a specific audio components

widemediaphotography

Active Member
Joined
Jul 2, 2023
Messages
128
Likes
33
I would like to propose a question or an exercise that I do not know is actually feasible.
Suppose we have 2 DACs, (A and B)
A with very good measures and musically neutral,
B defined by people with golden ears as an audiophile masterpiece, without having any discount in the measurements (there are many mythological devices and that's it)

What would it take to make DAC A "sound" exactly like B?
What Hardare system and Software could be adopted?

I imagine that by sending a series of known signals to the second DAC and acquiring all the measurements we can perfectly understand how to modify its A to make it as similar as possible via software. When I say as similar as possible, I am also referring to the timbre that depends on the higher-order harmonics in an audio signal. What tools do we have in place to do this?

In other words how do replicate exactly the sound of one specific audio components (DAC/Amplifier, etc.etc.)?

Thank
 

Purité Audio

Master Contributor
Industry Insider
Barrowmaster
Forum Donor
Joined
Feb 29, 2016
Messages
9,606
Likes
13,280
Location
London
Two dacs both with very good measurements will already sound like each other.
Keith
 

Timcognito

Major Contributor
Forum Donor
Joined
Jun 28, 2021
Messages
3,815
Likes
14,265
Location
NorCal
In electronics perfect output matches input, that's it. Don't what to make you feel bad... waist of time and time is life's most precious commodity. Unwatching now. :rolleyes:
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
3,176
Likes
4,197
If you had a very-good DAC and a very-bad DAC, plus a good interface/ADC, you could record the bad DAC (with Audacity or any recording software) and play it through the good one.

But that's virtually impossible because it's hard to find a DAC that's NOT better than human hearing. (I had a noisy soundcard once that made noise when the hard drive was accessed.)

Most audio electronics is very good and any audible defect/difference is likely noise (hum, hiss, or whine in the background). You don't need "golden ears" to hear noise, but of course if you have hearing loss you may not hear it. ;) Or, you can over-drive an amplifier into clipping/distortion.

On the other hand, ALL speakers/headphones make a difference and room acoustics make a difference too.

See Audiophoolery and What is a blind ABX test?.
 

EERecordist

Active Member
Joined
Feb 24, 2024
Messages
169
Likes
227

There are only a few DAC chip makers, so DACs today will be similar.

You will see variations in the anti-aliasing filter, but there is not a lot of information in the very high frequencies to hear a difference in music.
 
Last edited:

Jim Taylor

Senior Member
Forum Donor
Joined
Apr 7, 2024
Messages
316
Likes
890
B defined by people with golden ears as an audiophile masterpiece,

IMO, people don't have "golden ears". That phrase is commonly used by scam artists. :(

Also, beware of the term "audiophile". Some people use it to mean "a fan of recorded music", but some people use it to mean "anti-science" or "anti-logic". You really need to dig deeper to understand how the user means it.

And as @DVDdoug said, it's difficult to find a DAC that doesn't perform below the threshold of human hearing.

Bottom line? Don't believe everything you hear. If you have doubts, arrange a double-blind test.



Jim
 

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,952
Likes
7,997
Location
San Francisco
In other words how do replicate exactly the sound of one specific audio components (DAC/Amplifier, etc.etc.)?
Let's take the premise at face value and assume that DAC B really does have a certain sound, and it's composed of various distortion modes and frequency response variations.

I'm going to approach this from a DSP point of view.

This is something that is done very commonly in the studio world, especially in the guitar world.

First of all, anything we want to replicate needs to be measured. If it can't be measured then we certainly can't figure out how to re-create the sound.

The next step would be to characterize all of the distortions the DAC can put out, at every input level (probably 100+ samples), and record the impulse responses (FIR). This is necessary because with non-linear gear the transfer function changes at every input level. So we need to take many measurements and interpolate them as little as possible.

Next, "simply" put together a DSP chain that implements these distortions via convolution. This is basically what guitar amp modeling software does, just usually with much more extreme variations in THD and frequency response. https://www.ikmultimedia.com/products/amplitube5/

There are other ways to do this. If you can accurately figure out what type of distortion the device puts out, you can simulate that algorithmically. Variety of Sound has a lot of analog-like algorithms that do this kind of thing.

Or, you can simulate the circuits themselves in detail via brute-force, like engineering simulations of circuits. This is how the Uhe DIVA synthesizer gets such a nice analog-like sound. Doing it this way avoids uncertainty and approximations, but also requires a lot of CPU. You would need a full PC to run this.

NB: If we are talking about golden ears describing the sound quality of something, but it doesn't show up in measurements, then the method for replicating the sound is much simpler. Just do what the golden ears do... sit down and use your imagination.
 
Last edited:

Cbdb2

Major Contributor
Joined
Sep 8, 2019
Messages
1,645
Likes
1,630
Location
Vancouver

There are only a few DAC chip makers, so DACs today will be similar.
This is the answer. Interesting that there staring to emulate the circuit using virtual circuit components in a spice program and then run the signal thru the emulation real time. Any one who's used spice knows the processing power needed for this.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
21,240
Likes
38,628
OP
W

widemediaphotography

Active Member
Joined
Jul 2, 2023
Messages
128
Likes
33
In the meantime, I would like to thank the authors of the last answers, because they have placed themselves in the position of science in answering my curious question.

I have a D300 smsl DAC that reviews and measurements will confirm as being very good and from my tests I find it particularly neutral and quiet with very, very good channel separation.
However, the DACs don't all sound the same and depending on the setting. My DAC sounds different too, but that's because in its setup you can choose about ten presets and colors (Linear Fast, etc., Tube 1, 2, 3, etc, etc.), which do nothing but modify the waveform and/or add some EQ. It's just nuances, many of which I struggle to tell the difference using a HIFIMAN HE1000 Stealth headphones and a Violectric V222 amp; A setup chosen because it expresses a clear and open sound and above all very neutral.

Therefore, remaining in the sphere of science in which what is measured is heard and not vice versa, why shouldn't there be DACs or amplifiers that sound different?
Example tested on my setup, just replace my DAC and my headphone amplifier with the HIFIMAN EF400 which is also an R2R DAC, and the difference is definitely audible, in fact you realize that it is quite "colorful" without having to make any measurements.
All devices sold for audophile owls rely on their ability to distort the waveform and convince the listener that this is elitist perfection. On the other hand, if it were not true, there would be no HARMAN and RIAA curves, which change the sound of the equipment.

I already knew about the possibility of being able to operate via sw and in real time, but I missed the sw and the possibility of specific hardware. That is why I have asked the question in a general way.

My intent is to beat up the usual "golden-eared audiophiles" and reproduce with my DAC and their amplifier the exact same sound emitted by their so-called esoteric DACs, paid $10,000, which if it is true that they emit different sounds it is only because they distort the signal, changing the waveform. In addition, I would like to demonstrate analytically that such a DAC will never be able to reproduce a neutral sound.

Thank you again for your participation and attention to this discussion. :)
 

hifiole

Member
Joined
May 30, 2024
Messages
6
Likes
3
Bob Carver once made a transistor amplifier sound like a tube amplifier a long time ago, he had 48 hours to make the two amplifiers sound the same and he succeeded, you can read about it here

Modern DACs have very similar measurements and with values far below what we can hear, their performance is close to perfect and DACs should sound similar.

But do they? A "golden ear" design is typically a combination of design optimisation, attenuation and component selection to achieve a specific sound signature.

Whether there really is a difference between a ‘golden ear’ DAC and a more standard and probably much cheaper DAC can only be decided by the ear, I don't think there are any significant differences in the measurements between the two, also because measurements can only measure a small part of the listening experience, mainly what seems annoying to our ear, such as noise, strong deviations in frequency response, distortion, measuring what matters most for a listening experience can be difficult to impossible to measure it could be soundstage/image, tone/sound, rhythm, fidelity.

But according to ASrR, there are no differences here.
So I think you just have to believe Bob Carver, if the measurements are the same and the ability to deliver the necessary power is present then there is no difference in sound.

If the two devices have high SINAD , they will sound the same but without a sound signature.

If SINAD is poor, there will be a sound signature that a skilled technician like Bob Caver can copy , By changing data and thus get two devices that sound the same but with a sound signature.
 
Last edited:

Jim Taylor

Senior Member
Forum Donor
Joined
Apr 7, 2024
Messages
316
Likes
890
Whether there really is a difference between a ‘golden ear’ DAC and a more standard and probably much cheaper DAC can only be decided by the ear,

That is absolutely correct! The test "by ear" is called a Double Blind Test, and is discussed here.

... because measurements can only measure a small part of the listening experience, mainly what seems annoying to our ear, such as noise, strong deviations in frequency response, distortion, measuring what matters most for a listening experience can be difficult to impossible to measure it could be soundstage/image, tone/sound, rhythm, fidelity.

Sounds like perfect subject matter for a double-blind listening test! If you can identify it blind, it exists. If you can't, then it doesn't. :)

Jim
 

hifiole

Member
Joined
May 30, 2024
Messages
6
Likes
3
That is absolutely correct! The test "by ear" is called a Double Blind Test, and is discussed here.



Sounds like perfect subject matter for a double-blind listening test! If you can identify it blind, it exists. If you can't, then it doesn't. :)

Jim
I basically believe that a hifi blind test is more a test of the test itself than a test of the actual hifi object.

As a reason to use blind tests Amir mentions drug and food testing, in both cases there is a constant, the drug, and a variable, the human, if the test has enough participants, it is a reasonable marker no matter what the test ends up with.

A hi-fi test cannot be compared to this, it also contains a constant, the test object, but many variables besides the human, to name a few, the hifi-system used, the experience of the listener, the choice of music, the acoustics of the environment etc. and the test typically has few participants.

However, if the test gives a positive result, there is a good chance that it is correct.
On the other hand, if the test is negative, there is no guarantee that the result is correct due to the many variables.

Whatever the result is, it will always be tied to the test conditions and it will be very difficult for another independent group to repeat the experiment, which can easily be done in the case of medicine and food and this double-checking by an independent group is an important part of science.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
21,240
Likes
38,628
I basically believe that a hifi blind test is more a test of the test itself than a test of the actual hifi object.

As a reason to use blind tests Amir mentions drug and food testing, in both cases there is a constant, the drug, and a variable, the human, if the test has enough participants, it is a reasonable marker no matter what the test ends up with.

A hi-fi test cannot be compared to this, it also contains a constant, the test object, but many variables besides the human, to name a few, the hifi-system used, the experience of the listener, the choice of music, the acoustics of the environment etc. and the test typically has few participants.

However, if the test gives a positive result, there is a good chance that it is correct.
On the other hand, if the test is negative, there is no guarantee that the result is correct due to the many variables.

Whatever the result is, it will always be tied to the test conditions and it will be very difficult for another independent group to repeat the experiment, which can easily be done in the case of medicine and food and this double-checking by an independent group is an important part of science.

This is too feeble minded to be serious. Audio testing has too many variables, but drug testing doesn't? Like age, sex, other health variables, the jobs people do during the test, how hard they work or not, how rich they are or not, how stressful life during all the long test period is and in a myriad of different ways for each and every test participant. But a confined audio test has too many variables?? :p

Of course you also show a lack of understanding about statistics along with missing the reproducibility "crisis" in medical science. You aren't half as clever as you think you are, and I won't be calling you Shirley.
 

hifiole

Member
Joined
May 30, 2024
Messages
6
Likes
3

Dette er for svagt sindet til at være alvorligt. Lydtest har for mange variabler, men lægemiddeltest gør det ikke? Ligesom alder, køn, andre sundhedsvariabler, de job, folk udfører under testen, hvor hårdt de arbejder eller ej, hvor rige de er eller ej, hvor stressende livet i hele den lange testperiode er og på et utal af forskellige måder for hver enkelt testdeltager. Men en begrænset lydtest har for mange variabler?? :p

Selvfølgelig viser du også en mangel på forståelse for statistik sammen med at gå glip af reproducerbarhedskrisen i lægevidenskaben. Du er ikke halvt så klog, som du tror, du er, og jeg vil ikke kalde dig Shirley.
Instead of just pissing me off, I think you should answer some of the issues I raise.
 
Top Bottom