The famous AB test between amplifiers

[terryforsythe]

Hi everyone
I have read a lot on the web that if we match the outputs of the amplifiers in an AB test we would not be able to find any auditory differences. Let's say we choose a Carver 360 vs Topping LA90 amplifier, would the results be the same? Does anyone have a different opinion?

My opinion is that differences can be found in an AB test, but not with the traditional method of fast switching. In fact, I would say that if we were to do the tests in this way, even if we listen to one in stereo and the other in mono, we still wouldn't hear any differences.

It depends.

With modern solid state amplifiers, once you get above the budget class, essentially there is not an audible difference unless you put your ear next to the tweeter to listen to the noise level.

Budget class D amplifiers, though, oftentimes have a frequency response that is load dependent, and thus may sound different than higher quality amplifiers.

Tube amps sound different than solid state amps, at least the ones I have heard. And, different tube amps sound differently.

Long, long ago I had a Carver M1 amplifer. Carver tuned it to sound like a tube amplifier (tuning of the harmonic distortion characteristics I suspect). It definitely sounded different than other solid state amplifiers around at the time.

 

[kemmler3D]

I have seen 96 as the cut off point for "inaudible" mentioned several times. That's what I have been using at the cut point when thinking about things. This would be in the context of A/B testing.

Using an a/b switch and blinded, I was able to pick out differences between 88 and 106, volume matched, and pick the 106 sinad amp as having less noise (to my ear). But the 88 had load dependency feeding into 8 ohm speakers raising the high end (which is likely what I heard as more noise).

I appreciate the time you are taking here, thanks!

96dB SINAD is the equivalent to 16 bits of dynamic range, which is what CD has. True inaudibility varies depending on what you're doing with the signal... 96dB is mentioned a lot because of 16-bit digital audio, if you have more than that, you are transparent to 16-bit sources.

A lot of stuff becomes effectively inaudible when it's only 60dB below the signal, stuff that is 80dB down below music is generally inaudible, and then beyond that you get into special cases where you are listening to quiet portions of a recording with lots of gain, things like that.

I would bet good money that what you heard was load dependency and not noise or distortion per se. A 1-2 dB change in the treble is not super difficult to notice, but changes in noise or distortion that far below the music is damn hard to notice.

 

[MAB]

Hi everyone
I have read a lot on the web that if we match the outputs of the amplifiers in an AB test we would not be able to find any auditory differences. Let's say we choose a Carver 360 vs Topping LA90 amplifier, would the results be the same? Does anyone have a different opinion?

My opinion is that differences can be found in an AB test, but not with the traditional method of fast switching. In fact, I would say that if we were to do the tests in this way, even if we listen to one in stereo and the other in mono, we still wouldn't hear any differences.

Thanks for your opinion.
Perhaps you can conduct a scientifically controlled study that actually tests your hypotheses, and tell us the result.

One hypothesis appears to be that using fast-switching will prevent listeners from hearing differences, even between stereo and mono reproduction. Please test this ASAP and get back with the results. Should be really easy to demonstrate.

The other hypothesis is that a Carver and a Topping amp sound the same. The Carver has a distinct frequency response variation, likely large enough that the Carver is distinguishable from other amps under properly controlled tests. I am uninterested in opinions, just the results of proper tests. Please generate results, null or otherwise.

In the mean time, read one of the classic tests:

https://americanradiohistory.com/Archive-HiFI-Stereo/80s/HiFi-Stereo-Review-1987-01.pdf

It's on page 78. The rest of the magazine is for enjoyment.

And perhaps pay the fee and download this survey paper.

Actual Double Blind Studies

I have always been curious about 'training' for these tests. I saw a visual test that used an aerial pic of a full Rose Bowl, and instantaneous ABX testing was used to compare two pics and seeing if people could notice the difference between them. The subjects couldn't do it. Then, the tester...

www.audiosciencereview.com

The only amps that were distinguishable had frequency response anomalies, similar to the Carver.
Reading up on the AES studies will save you a ton of time and effort. ABX testing is not really that fun, but does provide truthful investigation as to what is relevant and what is not. And will show many of our 'opinions' are expensive distractions.

 

[Pareto Pragmatic]

96dB SINAD is the equivalent to 16 bits of dynamic range, which is what CD has. True inaudibility varies depending on what you're doing with the signal... 96dB is mentioned a lot because of 16-bit digital audio, if you have more than that, you are transparent to 16-bit sources.

A lot of stuff becomes effectively inaudible when it's only 60dB below the signal, stuff that is 80dB down below music is generally inaudible, and then beyond that you get into special cases where you are listening to quiet portions of a recording with lots of gain, things like that.

I am almost entirely focused on making what I have measure better. So I rarely think in terms beyond that. It takes a while for things I have not used and focused on to stick. You are nailing down a lot of vague knowledge I had floating around in my head, which I do appreciate.

 

[antcollinet]

But the 88 had load dependency feeding into 8 ohm speakers raising the high end (which is likely what I heard as more noise).

That is possible - but then you are not hearing the Sinad difference, you are hearing frequency response.

 

[Cbdb2]

Well, these two are not equally transparent, just check the frequency response of the Carver:
View attachment 390065
This might be audible. Also with an efficient speaker, the 50/60 Hz hum may be audible if you turn up the volume... You would not even need a signal to tell them apart. And the 44 SINAD certainly comes into the realm of audible as well.

Is that into a resistive load? Than Speaker impedance might make the freq response even more wonky. Tube amps. This is what people actually hear as a difference with tube amps that have low distortion.

 

[Sokel]

Real life experience is sometimes a good way for peace of mind.
It's all about conditions and noise keeping in check.

Listening to 100+ dB sensitivity horms at 4 m distance with outputs matched to 0.1mV at 500Hz is nearly impossible for a bunch of people to distinguish a Purifi from a pair of 6W hand made Japanese SET amps.

And that for both casual 70dB SPL (A) average and 90+dB SPL (A) listening at various works.
I'm sure that if I was about to measure them SET would be in the 50s or lower in terms of SINAD (it's distortion component,not noise) .

It rapidly becomes meaningful as long as the conditions are right.

 

[Blumlein 88]

After level matching with electronics frequency response is far and away the most common reason you hear a real difference.

 

[END WAY]

Of course, SOME amplifiers ARE defective or inferior...


"Fast switching" refers to quickly (or instantly) switching between A and B. You could listen to A for five minutes (or as long as you like) and then instantly switch to B and then listen to B as long as you like. If it takes several minutes to disconnect one amplifier and connect the other, our auditory memory isn't as good. If you wait until the next day it gets even more difficult to reliably hear a difference.

In an ABX test, the listener is (theoretically) allowed to listen as long as they wish, and switch as often as they wish (between A, B, and X) before making a conclusion. And the ABX test has to be run several times to get a statistically valid result. (If you run the test just once, you have 50/50 chance of "guessing" X correctly.)

If there is a difference or defect with modern solid state amplifiers, it's usually noise. Frequency response and distortion are almost always better than human hearing unless it's over-driven into clipping (distortion). A good tube amp will be the same but it's harder and more expensive to make a good tube amp. (McIntosh has made good tube amps since the tube days.)


In some cases that's true - I have a program called Wave Repair for "cleaning-up" digitized vinyl records. On of the click removal methods is to copy the good channel into the bad channel when the click is only in one channel, or when it's offset in time between left and right. When you lose stereo for a few milliseconds, you don't notice it!

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet and we are always letting ourselves be guided by measurements, to begin with if we listen with cheap speakers that have a high intermodulation distortion all that effort of low amplifier specifications goes to waste. That is why we must take into account that the speakers are decent.
But let's see a typical case, if we lower 0.25 decibels for example in 2000 Hz I am sure our brain will detect it but will we be able to visualize it in the response curve? I assure you not, it would be imperceptible and we could say it is within the margin of error.

These a/b tests exchanging immediately are absurd, I'll give you an example, listen to 10 Chinese people speaking and you'll have a hard time telling them apart but the native Chinese could tell them apart immediately, it's all a matter of adaptation
Let's go further, the mother and father of twins, I assure you, can find differences when their children talk to them, go to that house and try to identify the differences between twins in an a/b test immediately to see if you can

With this I mean that if you listen to your system many, many times and you know how your favorite song sounds and you change the amplifier for one of similar power but of lower or higher quality or different topologies, I assure you that you will be able to identify some differences, but go to a showroom with these a/b tests to see if you can identify small differences, I assure you everything will sound the same.

 

[MaxwellsEq]

I have seen 96 as the cut off point for "inaudible" mentioned several times

It's worth remembering SINAD contains elements of noise AND distortion. If all we cared about was distortion, 80 would probably be OK. We are much more sensitive to noise than we are to distortion, which is why a number like 96 may be referring to Signal to Noise Ratio (SNR), rather than distortion.

 

[Cbdb2]

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet and we are always letting ourselves be guided by measurements,

Not sure what that's supposed to mean but modern measurements are far more accurate than ears/brain.

But let's see a typical case, if we lower 0.25 decibels for example in 2000 Hz I am sure our brain will detect it but will we be able to visualize it in the response curve? I assure you not, it would be imperceptible and we could say it is within the margin of error.

That .25db is getting close to inperceptable but very easy to measure.

These a/b tests exchanging immediately are absurd, I'll give you an example, listen to 10 Chinese people speaking and you'll have a hard time telling them apart

That makes zero sense, peoples voices sound different no matter the language. And quick A/B comparisons are the only reliable ones. Long term are brains adjust our perception. This long term thing (like burn in) is a scam perpetuated by the snake oil merchants and there marketers, the audio rags, to get your imagination involved so you think things got better when they are the same. And audiophools dont like quick compares because it proves they can't hear a difference that there ego insists should be there after spending thousands on a cable//lifters/ground boxe etc.

With this I mean that if you listen to your system many, many times and you know how your favorite song sounds and you change the amplifier for one of similar power but of lower or higher quality or different topologies, I assure you that you will be able to identify some differences, but go to a showroom with these a/b tests to see if you can identify small differences, I assure you everything will sound the same.

And I assure you that you will hear a difference between two amps that are identical if you have some time between the two and if you think you payed twice as much for one of them. Go to a showroom and everything will sound the same because they sound the same, unless there's gross mismatches in freq. response, distortion etc, like SS to deliberately colored tube amps.
99.9% of changes when recording, editing, mixing and mastering music are done with quick A/B comparisons, no matter how much time is available, for a reason, there the most reliable.

 

[Petrushka]

If it takes hours, days, or weeks to adapt enough to hear a difference, that time would be better spent adapting so as to hear the music through the system. Musicians do this all the time.

I you can hear a 0.25 dB difference and consider that critical, you can never be happy with any real world system.

Being unhappy is a choice.

 

[MAB]

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet

I'm just gonna stop right there.
You may have the most advanced 'processing of our brain through our ears' on the planet, but I am not a bat, or a dolphin, or this guy:

Seriously, are you arguing that we have the best auditory processing of any species on the planet? Citations are needed. And if so, what aspect of our perception of sound is the best on the planet? I am eager for you to provide this.

 

[mhardy6647]

Seriously, are you arguing that we have the best auditory processing of any species on the planet? Citations are needed. And if so, what aspect of our perception of sound is the best on the planet? I am eager for you to provide this.

I mean... not to mention... you know... the aliens.

All About UFO's

I'm more concerned he's being used by them as a useful idiot to promote their psy op. I mean, yeah, you could hire this guy, show him just enough to get him worked up, then cut him loose and let him "blow the whistle" all he wants. And watch the tax dollars roll in for programs you neither have...

www.audiosciencereview.com

 

[Old_School_Brad]

• Inaudible noise?
• Inaudible distortion?
• Load independent? (flat response)
• Driven within power envelope?

If you can answer yes to these, the amplifiers will with all likelihood be indistinguishable in a proper blind test.

 

[Blumlein 88]

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet and we are always letting ourselves be guided by measurements, to begin with if we listen with cheap speakers that have a high intermodulation distortion all that effort of low amplifier specifications goes to waste. That is why we must take into account that the speakers are decent.
But let's see a typical case, if we lower 0.25 decibels for example in 2000 Hz I am sure our brain will detect it but will we be able to visualize it in the response curve? I assure you not, it would be imperceptible and we could say it is within the margin of error.

These a/b tests exchanging immediately are absurd, I'll give you an example, listen to 10 Chinese people speaking and you'll have a hard time telling them apart but the native Chinese could tell them apart immediately, it's all a matter of adaptation
Let's go further, the mother and father of twins, I assure you, can find differences when their children talk to them, go to that house and try to identify the differences between twins in an a/b test immediately to see if you can

With this I mean that if you listen to your system many, many times and you know how your favorite song sounds and you change the amplifier for one of similar power but of lower or higher quality or different topologies, I assure you that you will be able to identify some differences, but go to a showroom with these a/b tests to see if you can identify small differences, I assure you everything will sound the same.

Your little scenario sounds fine. Except it has been tested and it does not pass muster in any way, shape or form. Under no circumstances does it turn out to be true. NONE. ZERO. NADA. So bring us some counter evidence or accept the evidence which is deep and repeated many times.

The Chinese speakers thing and twins thing. If we switch rapidly and seamlessly, no one will know it ever happened. No amount of increased exposure will change that.

 

[Old_School_Brad]

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet and we are always letting ourselves be guided by measurements, to begin with if we listen with cheap speakers that have a high intermodulation distortion all that effort of low amplifier specifications goes to waste. That is why we must take into account that the speakers are decent.
But let's see a typical case, if we lower 0.25 decibels for example in 2000 Hz I am sure our brain will detect it but will we be able to visualize it in the response curve? I assure you not, it would be imperceptible and we could say it is within the margin of error.

These a/b tests exchanging immediately are absurd, I'll give you an example, listen to 10 Chinese people speaking and you'll have a hard time telling them apart but the native Chinese could tell them apart immediately, it's all a matter of adaptation
Let's go further, the mother and father of twins, I assure you, can find differences when their children talk to them, go to that house and try to identify the differences between twins in an a/b test immediately to see if you can

With this I mean that if you listen to your system many, many times and you know how your favorite song sounds and you change the amplifier for one of similar power but of lower or higher quality or different topologies, I assure you that you will be able to identify some differences, but go to a showroom with these a/b tests to see if you can identify small differences, I assure you everything will sound the same.

Those are two different kinds of memory you are describing in your example. I don't think it's really applicable in this context.

For audiotory differences our memory is extremely short. 2-4 sec. This is why quick switching is mandatory in a proper test if you are looking to reveal supposedly differences.

 

[EJ3]

Those are two different kinds of memory you are describing in your example. I don't think it's really applicable in this context.

For audiotory differences our memory is extremely short. 2-4 sec. This is why quick switching is mandatory in a proper test if you are looking to reveal supposedly differences.

If 2 SS amplifiers do not sound the same through the same speakers with quick switching in a double blind test: then either one or both have been 'tuned' for a certain "house" sound OR 1 or both are broken.

 

[voodooless]

We are underestimating the processing of our brain through our ears, we are the most advanced that exists on the planet and we are always letting ourselves be guided by measurements

We are (arguably) the most “advanced that exist on the planet”, exactly because we let us be guided by measurements. It helps us understand how the world works, or at the very least lead us to a working model of the observations.

We are the advanced not because our senses are exceptional, newsflash: they are mediocre at best, but because our ability to reason is generally very good…

What you are proposing is that our brain can do this:

 

[Geert]

These a/b tests exchanging immediately are absurd, I'll give you an example, listen to 10 Chinese people... <cut>

If you want to challenge the scientific consensus you need to do better than to come up with flawed analogies. Your Chinese people have the benefit of understanding other Chinese people, they don't recognise different people by the sound of their voice only. There's also no evidence that supports your claim.