Then just listen to your 16/44 CD's and thank the Gods you are one of that ever shrinking group who are a true disbeliever; Nothing could possibly be more realistic than a CD.
THx-RNMarsh
Last edited:
They certainly can do well, it's been very important for us to identify the direction of the sound of that twig snapping with speed and accuracy. That can require time processing in the single digit microsecond range.
Are you using the word "realistic" similarly to how you use the word "live" or do you mean something else?
Scott,
I will leave it to you to change the subject rather than admit your errors. Not willing to follow you down another rabbit hole.
Last edited:
The product of level of the data and sample rate must exceed a certain limit.
At 16bit levels, say a signal going from -32000 to +32000, enough intermediate sample values are available to reflect 1us and less time shifts (both absolute and interchannel). If you're down to three levels (+1, 0 ,-1) timing resolution cannot by any higher than 1/fs (26us for 44.1kHz) while preserving the reconstructed waveform.
44/16 has resolution issues for both amplitude and time once sample values are very low. 44/24 does better and will reach critical state at volumes probably below hearing threshold or buried deep in random noise. 96/24 even better....
At 16bit levels, say a signal going from -32000 to +32000, enough intermediate sample values are available to reflect 1us and less time shifts (both absolute and interchannel). If you're down to three levels (+1, 0 ,-1) timing resolution cannot by any higher than 1/fs (26us for 44.1kHz) while preserving the reconstructed waveform.
44/16 has resolution issues for both amplitude and time once sample values are very low. 44/24 does better and will reach critical state at volumes probably below hearing threshold or buried deep in random noise. 96/24 even better....
How astute of you to point it out now, though.
Sad, but we will probably always disagree about a few things. Hopefully, not too many.
Hans, please, try to read the followings with a open attitude.
First, nobody reject 44.1. RNM, JN, JC, Markw4, Bimo and I (with a lot of musician I know and, I think other people in the forum ) noticed that it is not perfect, listening to music. And it is just about attacks of some instruments (percussions).
It is not because we have a more performing hearing system, it is because we focus on this, and, may-be, not you.
We are not questioning the Nyquist theory, neither the Heisenberg's uncertainty principle applied to audio. That we question is the way our brains works to transform acoustic changes of pressure in the air into feelings. It seems it don't work exactly like the simplified system of a mike and an amplifier. And, may-be, it is not just our hearing system, but the way our speakers react to the electrical signals (acceleration). No sound stops like a brick wall at 20 000Hz in the nature.
There are two ways of thinking on this forum. The first seems to bring together people who start from a taught theory and try to make their personal experiences stick to it.
I personally think that it is a kind of unscientific belief, since the scientific approach begins with observation.
The second way is to listen first, then, try to find explanations. And the others objects that our observations (feelings) are flawed, subjects to external influences etc. (that they thing believes in audiophile B.S.)
We will not get out of it until the two parties make an effort to accept with an open mind, which may be true in the positions of the other party.
It is in the human nature to try to build better performing systems. 24/96 is more performing than 44.1.
0.00001% of harmonic distortion better than 0.1%.
Why to fight about this ?
I found funny the cbdb's remark that the equipment used in the studio are all "effect boxes" . Because it's true. As long as w'll have to use objects as full of faults as our loudspeakers, we will have to find, downstream, equipment as full as possible of inverse faults to compensate for them. Ad RNM says often, we have to look at the all chain. Not one element in isolation.
Good measured numbers is not the target. Our feelings, listening to music, is the one. Referring to the post juste above, who is naive ?
Last edited:
Amplitude modulation creates side bands. If we filter out the upper sideband, is the transient content of the signal the same? Do we hear the difference. It was pointed out that an audible difference was documented by someone. Preference was actually for filtered, but there was a difference heard.
I know of no reason for the developers of 44.1 to have conceived of the need to consider amplitude modulation products of certain percussion instruments back in the day.
That said, as we learn that the envelope modulation can force violation of nyquist with frequencies below the nyquist limit, we can either adapt to the new knowledge, or not.
With new understandings, we can now go down the ITD path if we want to, but again the math is there.
Jn
Last edited:
Yes, of course. Thank you for pointing it out in that manner.
We probably all know that 8-bit audio sounds pretty bad. The quantizing noise sounds more like gritty distortion to at least some humans.
Moreover, listening recently to a Chesky jazz sampler CD, it never seems possible to get low level little hand percussion instruments not to sound roughly chopped up in time (much like the early 8-bit synthesizers sound).
Actually, I believe he posted the information because some were saying mic's don't really go up to 20k. He showed otherwise, and mentioned someone else who consistently uses them.
I didn't see it as implying anything, just giving facts.
Jn
I filter all my amps. But it is for another purpose: avoid IM. As it depends of a lot of parameters, i do-it by ears. A compromise, as always.
Transients are my main purpose. That allow the presence and definition of the instruments, isolation, separation, positioning, call-it. Notice, it is always after 100khz (phase matters).
Extreme trebles brings more distortions than benefits. Slew rates is what matters. Extending the number of sampled points is not to increase the bandwidth, it is about sampling accurately the right level of all fast attacks. After this, we can reduce the bandwidth with a soft and natural descending curve . The difference will be the peak of the attack will not be lost somewhere between two samples.
One remark. One hit on a cymbal is not what matters. It is the average of the hits. If you begin a piece of music with a bass in solo, very defined, you can then drown it in the mass of instruments, you will always hear it with more definition than if it had been drowned from the start.
Last edited:
Now that you know why such things happen when you open your mouth, there should no longer be "annoyed and frustrated".
TT,one correction to your post.
You lumped me in with others as having noticed by listening to music that 44.1 is not perfect.
In my case, that is not correct. I have been perfectly content with the sound, as I have never noticed a difference other than an extended low end over vinyl because of groove spacing limitations.
My concern and understanding is with the fundamental understandings of how sampling can limit transient content should the nature of the transients(such as envelope modulation) cause an inadvertent violation of nyquist should the raw stream be digitized at 44.1.
It is an understanding that quickly developed with the help of the questions and objections of many posters here, and possibly would never have come to light without the input of others. To DPH, you mentioned the preference of the full presentation of a fully understood concept or hypothesis. That does not occur when the participants are hashing it out, only afterwards.
I find brainstorming in person to be far more manageable, as far less talking past one another occurs, and misunderstandings are far less frequent. Also, less frustration typical occurs.
I preside over such brainstorming groups at work, for problems which entail a very large range of special disciplines. It is far easier to get all to engage when in the same room, as opposed to a forum. Yet, this discussion does require several special disciplines, this thread essentially has them all.
Jn
You lumped me in with others as having noticed by listening to music that 44.1 is not perfect.
In my case, that is not correct. I have been perfectly content with the sound, as I have never noticed a difference other than an extended low end over vinyl because of groove spacing limitations.
My concern and understanding is with the fundamental understandings of how sampling can limit transient content should the nature of the transients(such as envelope modulation) cause an inadvertent violation of nyquist should the raw stream be digitized at 44.1.
It is an understanding that quickly developed with the help of the questions and objections of many posters here, and possibly would never have come to light without the input of others. To DPH, you mentioned the preference of the full presentation of a fully understood concept or hypothesis. That does not occur when the participants are hashing it out, only afterwards.
I find brainstorming in person to be far more manageable, as far less talking past one another occurs, and misunderstandings are far less frequent. Also, less frustration typical occurs.
I preside over such brainstorming groups at work, for problems which entail a very large range of special disciplines. It is far easier to get all to engage when in the same room, as opposed to a forum. Yet, this discussion does require several special disciplines, this thread essentially has them all.
Jn
Last edited:
I have understood that from your earlier remarks. In this case, you are neutral on the sound part of it . In fact, you are OK with the sound.
However, you do explain how and why what some others describe hearing is true.
THx-RNMarsh
Last edited:
🙂
It is a bit off my question... 44.1 limit is not what i asked to know. Though that must be known. I asked what rate is needed to reach the few usec referred to?
44.1 does not get there with 26us.
Can I figure it out? Sure.. Then why I asked it? I just wanted someone else to say it. Say what the number is needed to get to a few us. say it right out loud here in black and white. I've been saying it and its been danced around. Its someone else turn.
THx-RNMarsh
Last edited:
Over a decade ago I wrestled with the implications of a 2 uSec inverted bandwidth. But interchannel is not the same. As an engineer thinking temporally, I would want sampling sufficient to lock at the 1usec level. But I am confident that it can be done without such extreme sampling though.
I think the key is, the stream is not changing at that rate, we just have to keep the interchannel variance low enough (which I suspect already is at 44.1), and we have to make sure that any information used for ITD is not removed should the ITD factors we key to be a direct result of envelope modulation which violates nyquist.
Jn
Last edited:
- Status
- Not open for further replies.