Why don't you do a sound file that demonstrates the problem for us to listen to. Use music and auttenate and gain to your full harts content. Please describe the way you did the file in detail so we may repeat the process.
//
Wait. I put on my fire suit and give you my point of view. Personal, biased, flawed etc. like it should be.
First, except for manipulations of sound (DSP, amplification etc.) I do not find that the noise of a CD has ever been a problem.
It is the loss of definition of low-level sounds that is one of them, what I tried to illustrate with my little sound file that TNT does not seem to have noticed*.
Although this loss is not horrible to hear (the level being extremely low) it acts as a kind of expander. Low level signals have less and less presence as they drop in level. This drop in presence is added to that, natural, of our hearing.
I think that is what gives, in an insidious way, this impression of slight sterility (I can't find a better word).
It will be noted that this loss is more marked in the treble. Cymbals tend to disappear.
Which is a pity, because it is precisely on these instruments that the linearity of the decrease is most necessary.
The combination of the additional 8 bits (+ 48dB) and the doubled sampling frequency of 24-96 reject this phenomenon at a level which is now inaudible to my ears. Even when amplified at a level that could destroy both speakers and ears at 0dB.
Dithering allows ultra low levels to be audible, even when they are under the noise floor with 16 bits. It is better, but don't help a lot with the loss of definition I try to talk about. I think that it is will help nothing to a 24 bit file, because, near the noise floor, even the presence of absence of signal is not audible.
In most of the systems, for most of the consumers, the difference between 16X44.1 and 24X96 is not obvious. It is for people that listen in a critical way, with attention to technical defects, through a very focusing system.
I wonder why, each time somebody, here, try to put some words to communicate his feeling about this (sterile, reverb tails etc) he is entitled to the torrent of mud from the knights of the "I listen to nothing" and "my book tells us the opposite".
Just my two cents.
To J.C. it seems to me that you stay stuck to the first shock that digital has provided many who have experienced it the first time listening.
That you prefer the sound of vinyls as we prefer the titles of our adolescence and the emotion that accompanies them.
I invite you to record one of your favorite vinyls, if you have not already done so, in 24X96 and to compare them, after having synchronized them, in instant listening.
Besides the slight color differences that ADCs and DACs can introduce, you will see that it is impossible to tell the difference.
If you do the opposite, by cutting a 24X96 file on a vinyl, the difference will appear to you even before the first note. ;-)
*To TNT, Audacity is free, you can make your own experiments in any way possible. It's interesting.
Last edited:
Where is the sound file? I'm quite well familiar with DAW software. Make a file where you think the artefacts you talk about is audible... please? Your arw a sound engineer - right?
//
//
John Curl's Blowtorch preamplifier part III
The question was not to prove what is audible or not (each one his own) but to show the character of what could be audible. (your wife's face in a crowd)
As soon as we'll agree on the planning and my rates. Right?
Last edited:
Seems like we know what causes the improved sound. Did since early CD days when I did the Master tapes to compare to LP and CD.
So, lets break it down as to how the higher sampling rate contributes to the more accurate sound. And what more bits contributes to the more accurate sound.
The effect on the sound. Not electronically, or engineering wise but audio-wise. The engineering side is well understood and written about. I hear such an improvement that I stopped buying CD or any 16/44 transfers.
THx-RNMarsh
Last edited:
Still, you would most probably not catch which was which if someone played them for you without you knowing - given you poses a decent DAC. This is the most probable outcome of such a trial. Try again with someone doing it for you and report - please also report on how to acquire the used source material as I would like to repeat it for the given tracks. Or maybe this was how you did the examination? If so - congrats to good hearing!
//
//
No dac involved, if this comment was about my little file. Use your own source on the music you like. As long it contains the things that matter to you, human voices, percussion, attacks, held sounds, reverberations etc. I do not understand where you are coming from.
The first time I heard the Bop till you drop Ry cooder album (on vinyl), it was a shock for me. A clarity hitherto unknown to me, but also a strange sensation of a sort of emptiness. Also, I went out of my way to be able to experiment with this 3M 32 track machine.
Of course, I read everything I could on the principle of digital, I can not stand to use a machine which I do not know, at least, the principle of operation.
I will not tell the story, it would be too long and it was complicated.
Between the incessant drift of the machine, sensitive to the slightest temperature changes and other technical problems, I took my hours of sleep to experiment, in all possible ways, during the short time I had this machine available, the way in which this new technology interacted with sounds.
As we must do for any new equipment that a sound engineer has to test like a racing driver does laps to get the car in his hands.
Reading one another's comments, in reviews or discussions is of little use. Except if a criticism (positive or negative) is repeated and very largely concordant.
Hence the interest that I find in the multiplied exchanges of subjective impressions that seems, how could I understand why, to aggress lot of people here.
We must not only have our own idea, but also know how to easily recognize the character of each thing to be able to use it (or chose-it) wisely.
Theories and other measurement results say little about it. For me, they only serve to eliminate objects whose performance is clearly insufficient, to save time.
I experimented a lot with DACs, comparing the character of delta sigma, brick wall filters, oversampling etc. and the influence of their analog stages. As it seems to me that Mark4 does.
And I pay tribute to the incessant progress of our technologies.
This is why the aggressions of people like SYN08, which obviously did not make the hundredth of these efforts, but seems to live in an above ground world, amuse me enormously.
I do not think to convert to his religion in the near future and I feel a certain satisfaction to be part of the damned, whose every gesture deserves by default, and without examination, the wrath of his court.
I also find it very amusing to be treated of bodyguard by E.H.: With the 54 kg that I weighed before my wife's dishes, I did not even intimidate mosquitoes.
Last edited:
True. Perhaps a tendency that comes from the experimentation with sounds that goes along with studio recording of various popular genres. Its how one learns how to make certain sounds as heard on cleverly engineered recordings done by others. Partially analogous methods that may be used to help find a very clean and honest DAC sound. (Of course, measurements are needed too!)
Last edited:
Do you mean the DAC is like an effects generator, different effects improving different genres and/or recordings?
Just like Nyquist tells that the sample frequency Fs should be at least two times the highest signal frequency for correct reconstruction of the signal, the Widrow theorem tells that to prevent the signal to become mutilated leading to loss of signal, the added RMS value of the dithering noise should be at least 0.5Q, with Q being size of the quantisation step.
The RMS value of quantisation noise when digitizing a large enough signal is sqrt(1/12Q^2).
That means that when adding the proper dithering noise with an RMS value of 0.5Q, the SNR for a max Sinus shaped signal goes from (6.02W+1.76)dB to (6.02W+1.76-6.02)dB or for a 16 bit converter (W=16) from a theoretical SNR of 98.08 dB to 92.08 dB.
Since real SNR values are always a bit lower because of imperfections of the converter, a figure of 90dB will be what can be expected for properly dithered sinus shaped signals.
The signal with this added dithering, will stay intact for later reconstruction even when the signal is much smaller as Q, where a converter without dithering would stop working.
Even when drowned in the noise, recovery can later be done with band pass filters, but also up to a certain point by our auditory system.
Values of -10dB to -20dB below noise, depending on context and frequency, can be heard.
This is all perfectly in line with what Syn08 amongst others has mentioned before, leading to a 110dB dynamic range for a 44.1/16 processed signal, seemingly more than enough to enjoy music.
That was the reason for me to let JRiver convert a DXD master (352.8/24) and a few 192/24 files to several 44.1/16 dithered files and to compare them to the originals.
All files residing on my Windows 10 file system and played from there by JRiver through an asynchronous USB to ST Fiber converter, linked to my DAC/Audio system with ST Fiber.
So the chain to play was in all cases exactly the same, originated from the same master, played with exactly the same volume, the latter task also taken care of by JRiver,
I’m sure many won’t agree, but when playing at normal listening levels, I could not hear any difference at all.
Hans
The RMS value of quantisation noise when digitizing a large enough signal is sqrt(1/12Q^2).
That means that when adding the proper dithering noise with an RMS value of 0.5Q, the SNR for a max Sinus shaped signal goes from (6.02W+1.76)dB to (6.02W+1.76-6.02)dB or for a 16 bit converter (W=16) from a theoretical SNR of 98.08 dB to 92.08 dB.
Since real SNR values are always a bit lower because of imperfections of the converter, a figure of 90dB will be what can be expected for properly dithered sinus shaped signals.
The signal with this added dithering, will stay intact for later reconstruction even when the signal is much smaller as Q, where a converter without dithering would stop working.
Even when drowned in the noise, recovery can later be done with band pass filters, but also up to a certain point by our auditory system.
Values of -10dB to -20dB below noise, depending on context and frequency, can be heard.
This is all perfectly in line with what Syn08 amongst others has mentioned before, leading to a 110dB dynamic range for a 44.1/16 processed signal, seemingly more than enough to enjoy music.
That was the reason for me to let JRiver convert a DXD master (352.8/24) and a few 192/24 files to several 44.1/16 dithered files and to compare them to the originals.
All files residing on my Windows 10 file system and played from there by JRiver through an asynchronous USB to ST Fiber converter, linked to my DAC/Audio system with ST Fiber.
So the chain to play was in all cases exactly the same, originated from the same master, played with exactly the same volume, the latter task also taken care of by JRiver,
I’m sure many won’t agree, but when playing at normal listening levels, I could not hear any difference at all.
Hans
You don't look amused, more like sweating over the keyboard, but nevertheless, what do you think SYN08 was doing while you were trying hard to delude yourself?
Asking for a friend with an inquiring mind that wants to know
.No. I sure hope not anyway. Its just that FFTs and maybe time domain distortion residuals (to the extent we can make them accurately) don't quite tell us everything about how something sounds to a human. And they don't tell us which solutions to little problems will sound most proper (clean, honest, true, real, etc.) to humans (if there is more than one possible solution). Things like that still require listening and experimentation, IMHO (only!). Others may differ and be quite vocal about it, we'll see.
I've just found in the uncleaned mono pile a disc of Furtwangler conduction Menuhin. This has got to be worth a listen. Glorious early 50s mono!
I was thrown a bit by the genres/recordings comment. I believe in the "one size fits all (most 😉) solution to reproduction, in fact, that should be a good test of accuracy rather than the oft heard different systems for different genres/recordings me thinkst?
And by adopting this old cliché, you and your next of kin real or wannabe audio businessmen, instead of attempting to find a proper metric (if it exists), you open the door to charlatans. Good for you.
This is the audio "Secret Sauce" found by ear only, the holy grail of audio business for the last 50 years. Too bad it doesn't work in space communications, isn't it?
You removed me from your ignore list again? Why even bother putting me on that list when you can't keep your words. 🙄
You mean high fidelity DAC sound. That has already been in existence since the 90's.
I have a Furwangler doing Beethoven’s 9th at the 1954 Lucerne Festival. In immaculate condition and I listened to it only once. You are welcome to have it - PM me your address if you’d .ike it. Just sits on the shelf here.
Scottjoplin, if you don't believe that all the amps, all the integrated circuits, all the components we use sound (and measure) exactly the same, you have to accept that some people don't just choose their clothes for the insulation that they provide, but try to harmonize the colors.
Of course in an ideal world. But we all focus our attention to different details, our speakers are not the same, the recordings have all their character and bias. To re-use my previous image, it doesn't seem crazy to me someone want to adapt his outfit to the circumstances, although, like you, it's not my thing.
Conclusion, 16-44.1 seems perfect for you and you are an happy man.
All the previous has not so much interest when it is about what some can hear or not. The theory is nice, but limited at what it is.
As the laws of mechanics don't say much about our feelings behind the wheel of a car.
A nuance may-be too subtle to be understood by SYN08 master that takes what he have to believe in in books or other's opinions and, after, want to imposte-it to all others.
We are not listening to sinusoidal waves, but to very complicated musical signals that vary constantly both in frequencies content and envelope in time.
I'm totally sure that, if you compare (resampling them to 24X352.8 kHz ?) the two analog output contents of the two samples you compared, you will find differences in datas.
Now that you, or another guy, can hear or not those differences is another question, that only you can decide for yourself.
It is nevertheless curious this need to constantly refer to theories, books or the opinions of others, when it is so simple to trust our own feelings ...to feel aggressed when others do it, who only speak for themselves.
24-96 choice is not something AGAINST the laws we believe in of physics. It address something highly subjective: The human limits of audition, tastes, feelings. It is music, a lot more than the simple intelligibility of conversations between the moon and the earth during a spatial exploration.
Last edited:
Upsampling can not create new information, it can only add artifacts due to necessary compromises in implementation or deliberately added effects (the ken of the snake oil merchants).
- Status
- Not open for further replies.