Everyone should read this

Lots of good, well written info about audio, sampling, etc. here:

http://people.xiph.org/~xiphmont/demo/neil-young.html


(Ignore the link title - it doesn't really have anything to do with Neil.) 

there is a ton of stuff in that... I just skipped and skimmed it.... I copied the link to read it in depth later. 

thanks

Excellent article, thanks for sharing!

AT


Good article for sure.  Pretty complete, tho I didn't go over all the links and notes.
On the other hand, Rupert Neve says that higher sampling is the answer to digital not matching analog, and I'm loath to discount him.  Even at 80 his ears are golden.

If Rupert can't prove it with an ABX or double blind test (DBT), you should 100% absolutely, positively discount him. Or me.  Or Monty/xiphmont (who wrote the article). Or anyone else.

And don't just assume that someone who is a genius in the analog world has the same level of expertise in the digital world. Some of my favorite quotes from the article:

Misunderstandings of the mathematics, technology, and physiology arose in most of the conversations, often asserted by professionals who otherwise possessed significant audio expertise. 

And especially this one:

Sampling theory is often unintuitive without a signal processing background. It's not surprising most people, even brilliant PhDs in other fields, routinely misunderstand it. It's also not surprising many people don't even realize they have it wrong.

Rupert Neve says that higher sampling is the answer to digital not matching analog

I read an interview with Mr. Neve wherein he stated that he knows nothing about digital audio. That's entirely believable and understandable. I've known many old-timers who never felt the need to make the leap into the digital world. However, they also have the good sense not to make proclamations pertaining to subjects they admittedly know nothing about. Except maybe politics.

It all starts and ends with the human auditory system, the ultimate destination for everything we create, the final stage in the signal chain. Every serious student of this subject should really start there. So much about mixing becomes clearer in the context of sound perception, the mechanics of hearing, the many psychological aspects, even the role of evolution and hearing as a survival skill.

There is no substitute for study. Picking up the odd fact here and there on the internet is not going to make all the pieces fall into place. (Certainly not if you're soaking up unfiltered information from gearslutz. An actual quote: "the Nyquist theorem is wrong!" How do you even begin to rebut a statement like that?) 

-

-

-

-

In other words: the main point the guy is making is that you can't get meaningful results with a small number of tests. He's emphatically *not* saying you can't get meaningful results at all.

Oh man, that is completely not what he is saying. The conclusion you draw from the test is a fairly binary one. Is there or is there not a generally perceivable difference? His point is simply that you have to do a lot of legwork to be reasonably confident in your conclusion.

A typically meaningless response there. Whatever, people can make their own minds up. It is a shame that a thread drawing people's attention to something rigorous and informative has this woolly-headed misinformation on it though,

mike_mccue


I suggest you challenge yourself to a critical listen on something you know to have only a very minor difference and then see how you do.

The cool thing about Foobar2000 ABX is that you can export the results and share your experiences with others. Most people don't, and I think that is because the tabulated results most often reveal something that most people will only begrudgingly admit; ABX tests can teach us a lot about ourselves.

From a scientific proof perspective, you also have to be careful with how you do the test or the results are fairly meaningless. For instance if you decide beforehand you are going to do 20 test iterations, but stop at 12 (because you are getting the results you like), your results are not statistically valid.

Similarly if you do a test, and it gives you the "wrong" results (the ones you "don't want" for some reason) or not a high enough ratio of "correct" results, you can't just throw out those test results and/or repeat the same test until you get the "right" results.

So if you want to prove something, you have be very, very careful in how you go about setting up the test (so that there aren't any tipoffs unrelated to what you're testing for), and do a certain amount of work for statistically valid results (many iterations). There is much literature out there regarding this. 

This can make proving something (with a high degree of statistical certainty) tedious and difficult. But the reason it's tedious and difficult is, in part, what lends weight to the conclusions.

One of the main reasons pseudo-science is so popular is it allows people to pretend they are doing a test with conclusions that carry the weight of proper science - without having to bother with all those annoying rigorous scientific controls that force you to jump through all kinds of hoops. Of course the reality is that it's precisely those very rigorous controls that give scientific proof it's weight.


Having said all that, there's nothing wrong with using foobar (or another completely blind comparison tool) to get an idea how easy/hard it is to hear differences when you don't know what's what vs. when you do. And it can indeed be fun and enlightening.

... Is this Ethan Winer again?

Great toilet reading ... but like so many 'buyer-beware' articles, this doesn't really inspire music-creation for me.