@Goddard,
 
So - you are chastising folks for cluttering the forum by asking 'basic questions'?
 
Really?
 
There are lots of levels of knowledge in this forum, and while I am a decent keyboard player, and very good with computers,  I certainly am no expert in the details of audio fidelity, nor for that matter am I an expert in several areas of using Sonar.  
 
I would like to think I could ask a question without being judged for doing so.  This forum exists for folks to ask questions and have discussions.
 
All I can say is 'wow'.
 
Bob Bone
 
 
 
 
 

Well everyone has some area in which they are an expert in and I'm sure everyone else on this forum knows something else that you don't so I think the condescending attitude is uncalled for (although, yes, I've been guilty of that as well - I'm trying to be a good citizen these days).
 
Second, most people here understand how PCM data works, but expecting people to understand floating point numbers is not an easy thing to do since it is so different. What one need to grasp is that the 24 bits (effectively 25 due to the MSB always being 1), is scaled for each and every sample such that those 25 bits use all available bits to represent that one sample (which is why the MSB is always 1). It is like adjusting the gain of your signal path for the best possible SNR at each and every sample.
 
The amount that the data has been scaled by is stored in the exponent part (each 1 step increment to the exponent doubles the 25-bit range). This implies that as the signal approaches the zero crossing point, the resolution of those samples become finer (each step from one to the next possible step becomes smaller). On the other hand, as the signal level increases, the resolution becomes courser in order to not overflow the 25 bits. This is why floating point data is practically immune to overloads and your SNR stays constant no matter how low or high your overall signal level. The SNR for fixed point 25 bits is 150dB, but since in the case of float numbers the 25 bits are automatically scaled for best fit at each and every sample, the effective SNR becomes much much higher.

Goddard
I can see now that it was a mistake to even join this thread, and that I may have erred by overestimating the general comprehension level of many of the participants, but I was growing concerned due to that Noel's original post referred to that blog as a "great article" and as "well worth the read" and a number of people including Garrigus had then concurred that it was a good article/info, when it was quite obvious to me that it was written by a less-than-knowledgeable poser and riddled with misinfo and scaremongering and that there was a serious risk of folks less knowledgeable being taken in by it.

 
I personally did not see any scaremongering in that article and still assert it was reasonably well written for a broad audience without taking camps. I don't think it was intended to be an AES journal intended for the scientific community so some facts could be loosely stated and subject to interpretation :)
His premise that going above 96k had diminishing returns and the potential to actually screw up things more than help is not that far fetched to me. Maybe he could have provided more detailed evidence for that for the geeks but I don't think the article was written from that perspective or to go into DAC theory.  My take away is that it would be wise for anyone to think about the value of using a high sample rate before blindly going with the higher means better theory. A high quality converter at 44.1 will sound way better than a realtek 96k convertor :) I think the bottom line is every user has to test this for themselves with their specific hardware environment to see whether any benefits are tangible.
 
I understand the whole sample rate debate is a subject that people are passionate about please keep an open mind folks. This was not intended to start a flame war on sample rate theory but expose users to the fact that there are multiple pieces to the puzzle.

dubdisciple
Cry me a river.  i asked you a simple question and you go playing victim.  I can't say I'm particularly close to anyone on here enough to team up with them.  At most, maybe there are two in this thread clearly on opposite ends of the argument than you.  Some of us have independently been turned off by your posts without necessarily being on one side or the other but you are too hellbent on being "right" to pay attention to that.  Every time you are asked a question you respond with a different variation of "i'm right, you are wrong, here, read this article" .  I say good riddance.  There are plenty here willing to spew out the exact same info without being so combative about it. The fact that you were not nearly as belligerent to Noel, even when you disagreed shows that you are at least somewhat  aware of how combative you are being.  if you said the same things to Noel that you said to John and John, you would be making an involuntary exit.

Speaking for myself, I'm not that far on the opposite side of any argument. If you sift through all the long winded (and windy) posts herein, Goddard really only has one pertinent point; that Lavry's claim that sample rates over about 60khz are actually harmful seems to lack evidence.
 
I agree with him on that (though yer man Monty from Xiph has had a very good stab at a mathematical proof for the claim, it should be noted).
 
My issue all along has been what you're talking about above; the ridiculous show-boating.

Noel Borthwick [Cakewalk]
Also a high quality converter at 44.1 will sound way better than a realtek 96k convertor :)

Yeah, I raised an eyebrow at the notion that you could effectively test this stuff by tweaking the settings on a ropey motherboard converter. But by that point, my eyebrows were in danger of popping off the top of my head, so I thought I'd be best off not even getting into that one.

John
 
This is exactly right. The only thing I would add is in the subsonic zone. We can't hear it but we can feel it. 
 
Supersonic is not anything we need to worry about. We can neither hear it or feel it. 

 
A very good point, in terms of the pragmatics of this stuff, yes.

OK here is a simple test...
 
Create a project with the highest possible sample rate your setup supports. Then have some tracks that contain audio feed into different busses that each have various processing including compression, tube emulators, guitar effects like fuzz, distortion etc.
 
Now create a track with an ultrasonic signal in some way (I'm sure some plugins that generate test signals would be able to create ultrasonic signals). Now feed this HF signal into each of the busses, one at a time. Dollars to doughnuts your audible signal is noticeably altered in many cases.

We had a very thought provoking discussion at NAMM with Henry Juszkiewicz, Gibson CEO about how much more important the lower frequencies were rather than the HF that people obsess about. He's really clued in to this stuff and approaches it from a perspective that factors in the physiology of the human ear.
 

John T

John
 
This is exactly right. The only thing I would add is in the subsonic zone. We can't hear it but we can feel it. 
 
Supersonic is not anything we need to worry about. We can neither hear it or feel it. 

 
A very good point, in terms of the pragmatics of this stuff, yes.

Very few people would own speakers that can recreate subsonic frequencies to the extent that they can be "felt", so I wonder if it is even something worth worrying about at all?

and for the record, I fully apologize for repeatedly comparing Goddard to feminine hygiene products.  It is completely out of line and does not help matters. At some point I will look over the links he posted in an attempt to separate knowledge from abrasive personality.