John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Of course, the difference between the two of us is the size of our ego. While some may think mine is so huge that they wonder how I ever get through any doors, it is in fact quite small and cute, easy to carry, a lot like a Pokemon...

Since I am full time in the sound system business there is a 14' door at my shop. It almost fits all audio egos!
 

6) My overall conclusion is that the variability of the test conditions when compared to a person's usual listening situation often outweigh the differences which may exist.

This is probably a major consideration concerning various listening tests.
From my own experience, on my own stereo setup, I know I hear fairly subtle changes to various aspects of the sound – blind and sighted alike. However when listening to any other stereo setup, the overall sound signature of the other setup is so much different to me ears, that I doubt if I'd be able to detect even changes that cause gross changes to the sound.
 
To me, these DBT - ABX supporters are responsible for current catastrophic quality of recordings (except for classical music).

I have no idea whether DBT - ABX supporters are to blame, or not, however I find many of the current classical music recordings not as satisfactory as those from the "Golden Era". It seems like in the "Golden Era" recording engineers and producers paid attention to capture the ambiance atmosphere of live concerts. Most modern recordings capture excellent details by using multi-microphones, placed close to the musical instruments – but the overall sound is very far from attending live concert.
 
Hi,

Ah, sorry, I thought you were being serious about being able to distinguish things by ear alone. My mistake.

As we are back to bring out that old ABX chestnut...

ABX testing is presented by it's advocates as "scientific" and "reliable" method to determine the audibility or not of a given posited possible difference.

Going by the tests published by the most vocal advocates of this kind of testing so far we find:

1) The test is carried out without any controls, positive or negative, which are considered absolutely mandatory for scirentific testing.

In any field other than Audio the kind of tests published by the ABX Supporters (I normally have a stronger and more appropriate word here, but things being what they are I have to watch what I type, lest the censor striketh) would be torn to shreds on this account alone.

2) Many ABX tests, especially the ones receiving "Headliner" status, such as certain ones by Tom Nouisaine that the late "Audio" Magazine had the misfortune to publish, are carried out under conditions and in contexts that can be described, if we want to be very charitable as "p.e.e. poor".

They often are issued as "challenges" of the kind that say "prove to me what you think you hear". Even in near ideal conditions this alone contributes a significant psychological pressure which will in the test (which of course still misses any controls) act as significant randomising agent, that is it will create a predisposition towards a greater randomness, as may be reasonably expected (and has been illustrated in fields outside audio, such as Medicine with disturbing frequency).

To call such tests scientific or to present them as serious evidence for anything would get laughed out of the Mensa in any field but Audio.

3) Due to their presentation, aims and promotion, ABX tests are highly likely to attract individuals who are strongly biased about the topic being tested. It DOES NOT matter which way their biases run, either the Placebo effect or the Nocebo effect will act as strong randomising agent.

Again, should you advance such tests with sufficient background information in any scientific field except audio, the result would probably be that your research grants are cancelled (the commercial equivalent of this arguably already happened to the ABX Corporation, no wonder the guys who had their money in it are not pleased and like to rain on other peoples parades) ...

4) ABX Statistics are weighted severely towards avoiding "false positives" (Type A statistical error), to a degree that with the typically very small sample sizes a severe risk of "false negatives" (Type B statistical error) is incurred. In fact, the ABX statistical criteria used are generally severely inappropriate for the sample sizes present.

In addition, ABX proponents have a tendency to use the statistical device of a "lucky coin" to dismiss any results that despite all the prior efforts to return a "null" result fail to do so.

It should be noted that all of the above issues have been raised repeatedly in a range of pubications over the last 30 Years or so and the response from the ABX supporters has been to dismiss any criticism summarily without presenting evidence to actually counter this criticism (not that such could exist) and to "carry on" like before, putting one in mind of Captain Tobias Slater in the seminal "Wytchlocater" Skit in Dr. Terrible's House of Horrible...

"But we have this here ducking dtool..."

Ciao T
 
Most modern recordings capture excellent details by using multi-microphones, placed close to the musical instruments – but the overall sound is very far from attending live concert.

Yes, I agree. The modern recordings of classical music are usually 'over technologized'. Anyway, there are some exceptions, captured with less mikes and from longer distance.

Last time I was thinking about this 2 weeks ago when I heard a small acoustical jazz combo (4 instruments) playing at the open space square (Hradcanske namesti) in front of the Prague castle. Such kind o natural, full and rich sound is never heard from any audio chain.
 
Oh! ABX is not GOOD ENOUGH these days? What is better, in your opinion?

Not better, just different- ABX is one test format among many. There are lots of valid controlled listening protocols. See my recent article in Linear Audio for a review of several good ones.

My most recent test participation used a sorting protocol- you'll recall that Prof. Hawksford coded six files with three of them having a phase shift added. It was up to the listener to decide which three. Pano and I independently came to the same answer, which turned out to be correct. Not quite enough trials to achieve 95% significance (unless we do the dishonest trick of adding our results together), but that would just mean increasing the size from 6 to 8 files needing sorting. The point is, though, that it's another valid DBT format.
 
AX tech editor
Joined 2002
Paid Member
Last time I was thinking about this 2 weeks ago when I heard a small acoustical jazz combo (4 instruments) playing at the open space square (Hradcanske namesti) in front of the Prague castle.

slightly OT/
I've had a very impressive tour of the Prague Castle in rooms and areas off-limits for tourists. One of the perks of working for NATO.
You guys have so much culture that is only now going to be known to the rest of the world.
Hope I can return one time in the future.

jan
 
I find many of the current classical music recordings not as satisfactory as those from the "Golden Era". It seems like in the "Golden Era" recording engineers and producers paid attention to capture the ambiance atmosphere of live concerts. Most modern recordings capture excellent details by using multi-microphones, placed close to the musical instruments – but the overall sound is very far from attending live concert.

In America this is often blamed on orchestra costs and the requirements of technical perfection - small fixes to intonation or phrasing of a section can be made comparatively inexpensively - but I blame the influence of pop recording style. Folks who never hear live orchestral music think it *should* sound like you're standing in front of the conductor. And that the tiniest imperfection is a fatal flaw.

To my ear the bigger difference nowdays is that we have no more Barbarolli's, Beecham's, Reiner's, Munch's, Paray's, Dorati's etc. Plenty of good musicians today, sure, but how many of them actually *met* Brahms? Different world.

Thanks,
Chris
 
In America this is often blamed on orchestra costs and the requirements of technical perfection - Thanks,
Chris

Very much to the point, FWIW the conductor is no longer allowed to be a tyrant and force repeat rehersals, so principles are chosen on their ability to read once and play to rote perfection. I have this from principles in the BSO arguably one of the worlds better orchestras.
 
Administrator
Joined 2004
Paid Member
Don't any of these listening tests have immediate feedback? I've asked this before. If you have no feel for how you're doing, it's hard to get good at the test.

Just keeping a running score of right and wrong answers would be a help. Winning money would be even more motivation to get it right. :) But seeing your scores at the end of the test and not really remembering what you heard? Bah!
 
Don't any of these listening tests have immediate feedback? I've asked this before. If you have no feel for how you're doing, it's hard to get good at the test.

Just keeping a running score of right and wrong answers would be a help. Winning money would be even more motivation to get it right. :) But seeing your scores at the end of the test and not really remembering what you heard? Bah!

I agree!!! Without letting people practice ahead of time, you're just setting people up for failure. Especially when people unfamiliar with the equipment they are listening to.
 
If you want an 2x preference test of doubtful use, sure, no need to put in negative controls. Or eliminate them and increase the number of trials so that the participants can have a "fatigue" excuse.

"Doubtful use" is obviously wrong because significant results are just that.

Raising the number of trials would not help wrt to missing negative controls, but would help if you don´t want to use a positive control. But then you have to do several hundred trials and that is not a realistic choice.

@ pano,

yes, that is done during the training before the "real event" .

To include a result feedback in a test is not a good idea as every trial should be independent from any other trial and that might be influenced by any feedback.
 
Status
Not open for further replies.