Can you hear a difference in Sound between Audio Amplifiers?

[Steve81]

When Harman uses DBT to understand listener preference in the pursuit of profit, it makes sense to me. They're using the best data they can get for use in mass production. That's a far cry from 3 @ssholes sitting in a basement with a curtain.

That's more or less where I sit. Biased controlled testing is very necessary for professional / scientific research to determine things like audible thresholds, preferences, etc. This kind of work by Toole, Olive, et al. has been instrumental in driving the state of the art forward. Even as consumers, we would be wise to understand the implications, i.e. what a "good" amplifier/speaker should do, what specs are worth fretting over, etc.

On the other hand, we listen to our systems every single day with all of our biases fully in tact. Maybe a cheap Behringer measures as well as a much costlier ATI, and under blind test conditions, it sounds identical. However, if the Berry can't even compete with my son's Fisher Price stereo in terms of build quality and aesthetics, am I going to enjoy having it in my system? Probably not. Will it affect what I hear? Quite possibly. So am I better off buying the Behringer or the ATI, even if I "know" they will sound the same?

 

[fuzz092888]

That's more or less where I sit. Biased controlled testing is very necessary for professional / scientific research to determine things like audible thresholds, preferences, etc. This kind of work by Toole, Olive, et al. has been instrumental in driving the state of the art forward. Even as consumers, we would be wise to understand the implications, i.e. what a "good" amplifier/speaker should do, what specs are worth fretting over, etc.

On the other hand, we listen to our systems every single day with all of our biases fully in tact. Maybe a cheap Behringer measures as well as a much costlier ATI, and under blind test conditions, it sounds identical. However, if the Berry can't even compete with my son's Fisher Price stereo in terms of build quality and aesthetics, am I going to enjoy having it in my system? Probably not. Will if affect what I hear? Quite possibly. So am I better off buying the Behringer or the ATI, even if I "know" they will sound the same?

I'm all for fretting over specs up until I can't hear a difference anymore. Then I want pretty.

Exhibit A



I can't definitively say it sounds better than any other amp I own or have owned, but I enjoy the heck out of it.

 

[Steve81]

3 @ssholes ... with a curtain.

PS: This sounds like how you guys warm up for a Twister match

 

[Irvrobinson]

I'm all for fretting over specs up until I can't hear a difference anymore. Then I want pretty.

Exhibit A



I can't definitively say it sounds better than any other amp I own or have owned, but I enjoy the heck out of it.

The Threshold amps from that era were beautiful, inside and out. That "Stasis" circuitry is very similar to the Quad "current dumping" design, though on a far grander scale for the high power models. I almost bought an S550e, but ended up with Levinsons instead.

 

[fuzz092888]

The Threshold amps from that era were beautiful, inside and out. That "Stasis" circuitry is very similar to the Quad "current dumping" design, though on a far grander scale for the high power models. I almost bought an S550e, but ended up with Levinsons instead.

I've liked the S/300SE so much I'm still hoping to score an S/500(SE) or even better a pair of SA/12e or SA/1's.

 

[PENG]

However, if the Berry can't even compete with my son's Fisher Price stereo in terms of build quality and aesthetics, am I going to enjoy having it in my system? Probably not. Will if affect what I hear? Quite possibly. So am I better off buying the Behringer or the ATI, even if I "know" they will sound the same?

I would add that in the long run you may even get what you pay for going with ATI and other high quality amps because they are designed to last much longer and remain on specs. That's another important variable to consider when comparing newer and much older amps, the older amp may look new but depending on how it has been used or abused, it may not perform "like new".

 

[Irvrobinson]

That's another important variable to consider when comparing newer and much older amps, the older amp may look new but depending on how it has been used or abused, it may not perform "like new".

Very true, especially if their capacitors are aging, which is a common malady in all electronics older than 20 years.

 

[Steve81]

Irv – I can't disagree more with you. Because listener's attitudes, beliefs, and expectations play such a large role in their perceptions, blind testing is ESSENTIAL to determine whether people can actually discern subtle differences because of what they actually hear, as opposed to what they believe they hear.

I understand what Irv is saying, and I suspect some of it comes down to how the test is designed/implemented. Allotted time and finding material that highlights a difference are two important factors. Having a 1 hour session playing selections of Metallica, AC/DC, and Megadeth might not be the ideal way to pick out subtle differences. Having a dozen other people over who are also taking the test while contributing their own background noise (coughing, whispering, etc.) doesn't help either.

A more palatable test might look like this:
I give you two black boxes marked A and B, and hook them up to a switcher. The boxes are tamper proof so you can't peek inside, and of identical weight so you can't get any potential clues (relevant or not) that way. You have a month or two to decide if you prefer one over the other, in your home, with whatever you want to throw at them.

 

[Irvrobinson]

A more palatable test might look like this:
I give you two black boxes marked A and B, and hook them up to a switcher. The boxes are tamper proof so you can't peek inside, and of identical weight so you can't get any potential clues (relevant or not) that way. You have a month or two to decide if you prefer one over the other, in your home, with whatever you want to throw at them.

Awesome, but complicated and possibly expensive to implement.

 

[Steve81]

Awesome, but complicated and possibly expensive to implement.

That and it would take a while to collect a meaningful number of trials, as ideally you'd still collect multiple runs from each individual so you don't toss out the possibility of specific individuals having a preference.

 

[Irvrobinson]

That and it would take a while to collect a meaningful number of trials, as ideally you'd still collect multiple runs from each individual so you don't toss out the possibility of specific individuals having a preference.

Your technique is a variation on how I choose to buy electronics (unless they're cheap). I insist on taking home a sample and I live with it for a week. If I find I can listen for hours without complaint or fatigue, and it meets other build quality and ease-of-use requirements, I order one. If it gives me any sort of listening fatigue at all, it's out, and the audition is over. Over the past 25 years or so I've been amazed at how many amplifiers, pre-amps, DACs, and CD players can't pass these tests, though recently I haven't had any trouble with DACs. I think the latest ICs have completely leveled the DAC playing field.

I'm surprised that after all these years I still have opinions about Class A/B amplifiers, though the latest combination of SOTA IC op-amps for the first two stages and output transistors with awesome specs would lead one to believe that differences should be a thing of the past. I can't speak much to the Class D amps, as I have so little experience with them. Good thing I'm out of the market for the foreseeable future, so I don't have to worry about this stuff.

 

[Steve81]

That's the nice part of Internet Direct, i.e. a useful trial period. Only downside is in many cases the only thing people are comparing the new stuff with is the old crap they wanted to replace.

 

[Swerd]

I certainly agree that music selections could be important. While I agree that selections like Metallica, AC/DC, and Megadeth might not be the best, I don't know what good music selection might be. That seems to vary with the listener. Let the listener choose.

I have a problem with this part: "You have a month or two to decide if you prefer one over the other, in your home, with whatever you want to throw at them." Finding a preference assumes there is a difference. Instead, ask the listener only if A sounds different from B.

Here are further essential tests that would complicate things while doing a month-long A vs. B comparison:

1. The listener must also listen to negative controls to determine how often a listener reports hearing differences when there are none. Let the listener hear A vs. A, under blind conditions. How often does he report hearing a difference. Don't assume it will be zero, measure it and find out.
   
   
2. Similarly, the listener must also listen to positive controls to determine how often a listener reports hearing no differences when there are genuine differences. Again, it would be wrong to assume all listeners could always hear these real differences when you could measure it and find out. These positive control tests could be done by listening to the same musical passage with and without added white noise at various levels, such as 0, 2.5, 5, 7.5 & 10%. The positive controls play an important role in any listening test because it could establish a threshold level above which most listeners can report hearing differences.

With these negative and positive controls done at the same time, the results of listening to amps A vs. B might actually take on some meaning. Without them, the simple version of listening to A vs. B would mean very little.

While I'm not saying I don't trust listeners, how do you make certain the A boxes, B boxes, as well as the experimental controls are not tampered with during the month or two? That's a long time period.

And it goes without saying, that enough people (about 25 to 50) would have to be tested for any such test to have minimum statistical significance.

All this clearly is not trivial. But without it, all claims about what people can and cannot hear are meaningless.

 

[Steve81]

I have a problem with this part: "You have a month or two to decide if you prefer one over the other, in your home, with whatever you want to throw at them." Finding a preference assumes there is a difference. Instead, ask the listener only if A sounds different from B.

Agree, though that's why I included "if" in the statement.

With these negative and positive controls done at the same time, the results of listening to amps A vs. B might actually take on some meaning. Without them, the simple version of listening to A vs. B would mean very little.

I've got no beef with that either

While I'm not saying I don't trust listeners, how do you make certain the A boxes, B boxes, as well as the experimental controls are not tampered with during the month or two? That's a long time period.

If you wanted to be super-paranoid, plant a secret video camera. Barring that, just build the boxes in such a way that any attempts at tampering are evident.

All this clearly is not trivial. But without it, all claims about what people can and cannot hear are meaningless.

Indeed, but that's the problem. The kind of biased-controlled test were talking about requires resources beyond your average AV club, or even a publication like AH. So when do we start?

 

[Swerd]

Indeed, but that's the problem. The kind of biased-controlled test were talking about requires resources beyond your average AV club, or even a publication like AH. So when do we start?

It probably takes more motivation than resources. And I'm certainly not volunteering .

But I do believe that meaningful results can be found in an afternoon, a whole lot less time than a month or two. I've done that in the past, participating in a blind test of different crossover capacitors with about 40 other people.

I can understand if Irv prefers a longer time frame to decide on items he might purchase, but I don't think its needed to establish whether or not listeners can hear something.

 

[Steve81]

But I do believe that meaningful results can be found in an afternoon, a whole lot less time than a month or two.

It's certainly possible that they can. Still, for this kind of testing, a longer period of time helps, both in allowing a listener to utilize a wide range of material, and for alleviating one of the other concerns usually levied at blind testing: stress.

 

[Irvrobinson]

But I do believe that meaningful results can be found in an afternoon, a whole lot less time than a month or two.

I believe meaningful results can be achieved in an afternoon for speakers, but not for electronics. Except in rare cases I can't hear any differences between electronics on quick comparisons, but I do sometimes experience differences in listener fatigue, and in some cases that takes much more than an hour. And I like to do it over multiple days to rule out anomalies with my own perception as much as possible.

 

[Irvrobinson]

It's certainly possible that they can. Still, for this kind of testing, a longer period of time helps, both in allowing a listener to utilize a wide range of material, and for alleviating one of the other concerns usually levied at blind testing: stress.

In my experience the problem is not stress, it's simple guessing, especially with people who think they're golden eared. Guessing means garbage-in/garbage-out.

 

[Irvrobinson]

In my experience the problem is not stress, it's simple guessing, especially with people who think they're golden eared. Guessing means garbage-in/garbage-out.

And, yes, Swerd, I know that the statistical analysis filters out garbage, but it also points to a flawed methodology, which is why I don't like blind testing for audio.

 

[Swerd]

And, yes, Swerd, I know that the statistical analysis filters out garbage, but it also points to a flawed methodology, which is why I don't like blind testing for audio.

The source of the garbage is not the flawed methodology but the humans and their variable perception. That's why a large N is needed. All the negative and positive control results become a measure of how flawed or how sensitive the methodology is.

Stress, in my experience, comes from having to listen and focus on listening tests with only subtle differences. It becomes boring & difficult to keep interest up. People start guessing just to move on with the test. The tests have to mix it up, varying the difficulty from subtle differences to easily detected large differences.

The negative controls also measure guessing. They determine the "false positive" reporting rate. As listeners get bored or fatigued, and guess more, their false positive rate goes up.