Do all amplifiers sound the same? Level matched listening test

[MaxBuck]

H2 (deuterium) is an isotope of hydrogen.

H2 is also hydrogen gas.

 

[Chr1]

(... It is also a high performance supercharged motorcycle made by Kawasaki.)

 

[Thomas_A]

We are getting more votes on “I can hear a difference” without posting a valid ABX test report. Please note that such votes are not considered as trustworthy.

I've tried today but it is very difficult or impossible; the beat at 11.9-12 s might have something. I first managed 12/16 but then failed miserably on repeating to get better score. Tried again but got tired listening to the beat. Perhaps I will give it a last try another day.

 

[mixsit]

It is trivial to create test cases where no difference is heard, even when we know audible differences can exist. To wit, in my last job, we commissioned a test of 64 kbits/sec lossy codec against CD rip. The independent lab that we hired ran the double blind tests, showing that some 90% of the population couldn't tell the difference. We got lucky as the lab used classical music which was very easy to encode. If they had picked other material, the result would have been the exact opposite.

Key here is that one has to understand the system's weakness and psychoacoustics, and choose content that pushes that limit of audibility. We have that for codecs and speaker/room testing because work was put in to find such revealing tracks. For things like amplifiers, we don't have such critical tracks but that doesn't mean you just throw one piece of music at it and call it done.

I realise this is quite late in asking, but I was hoping to know about the music which is more difficult to encode? Actually perhaps better would be 'why? Thank you Sir!

 

[krabapple]

I realise this is quite late in asking, but I was hoping to know about the music which is more difficult to encode? Actually perhaps better would be 'why? Thank you Sir!

as to what, look up samples that are used to stress-test lossy codecs e.g 'castanets'. Or, see jj:

Topic: most difficult genre of music to mp3 encode?

The answer is 'some of all kinds'.

Solo Harpsichord (not just the SQAM thing) is hard.
Hammered Dulcimer (the Cooder thing) is ((*&* hard.
Suzanne Vega is hard.
Percussion music (the asian variety) is sometimes hard.
Low-pitched, open-vocal tract spoken voice can be hard. (German Male speech from SQAM)
"Glockenspiel and Drums" is pretty brutal (but made deliberately)

Broadband classical generally isn't. But that's only "generally".

 

[mixsit]

as to what, look up samples that are used to stress-test lossy codecs e.g 'castanets'. Or, see jj:

Topic: most difficult genre of music to mp3 encode?

Wow thank you. I've just cracked the top of the page but I think this is going to be a good read!

 

[danadam]

Wow thank you. I've just cracked the top of the page but I think this is going to be a good read!

In Personal Blind Listening Test of Bluetooth codecs thread there are links to 2 sets of samples:

• Kamedo2: https://hydrogenaudio.org/index.php/topic,98003.msg815018.html
• IgorC: https://drive.google.com/file/d/0ByvUr-pp6BuUSnlwUG1WNkZzeDA/view?usp=sharing

Many samples in Kamedo2's set are from https://www.rarewares.org/test_samples/

It seems the google drive link requires some access permissions

There are 3 additional samples in Vorbis aoTuV b5 @ q5 testing samples thread.

 

[pma]

Still waiting for abx reports from those who just checked the option they could hear a difference.

 

[JSmith]

I listened a few times before running the Foobar ABX test and was convinced I could hear a very, very slight difference. I felt amp_Y had a slightly different presentation of vocal edges and decays. To be clear, they sound so very similar

Agree, same here... yet I couldn't get a decent result in a Foobar ABX either. If there is a difference, it's so minor that one can't reliably identify it using this test track... which I don't think is the best for this test.


JSmith

 

[pma]

In my experience it is always the same case. We are quite sure we can hear a slight difference, which is then not confirmed by the ABX. Then we tend to blame it on the test track or the ABX method. With another test track it is again the same case. OK, we may say that with the test track used or equipment used we could or could not reliably tell the difference.

IME I never found the properly level matched and time aligned test that would reveal sound difference (with music) solely based on harmonic distortion difference. Contrary to that, quite small differences in amplitude frequency response were audible. Which is not the case in this test.

 

[Sokel]

In my experience it is always the same case. We are quite sure we can hear a slight difference, which is then not confirmed by the ABX. Then we tend to blame it on the test track or the ABX method. With another test track it is again the same case. OK, we may say that with the test track used or equipment used we could or could not reliably tell the difference.

IME I never found the properly level matched and time aligned test that would reveal sound difference (with music) solely based on harmonic distortion difference. Contrary to that, quite small differences in amplitude frequency response were audible. Which is not the case in this test.

In other words is better to have an amp with a little elevated distortion than a load dependent one.
Makes sense.

 

[Thomas_A]

I have not tried again, but think it is very difficult if not impossible. My first try that I made some days ago triggered me to try repeat it, but I failed. It was the beat around 11.9-12 s that caught my interest. I have no idea if there is any difference in measurements at this particular point, but if it is identical, I will surely not make another attempt.

 

[30 Ounce]

Why do the results add up to 108.4%?

 

[pma]

Because multiple answer options were allowed. Some members checked more than one answer.

 

[pma]

OK, the poll has been closed, with this result:

First, thanks to everyone who has contributed to the thread and voted in the poll.

I am missing ABX log reports from those who voted "I can hear a difference". Sadly, such votes without the DBT proof are quite meaningless, at least to my exploration.

Now, the disclosure. Amp X is my class AB A250W amplifier, 2 x 250W/4ohm rated continuous power. Amp Y is Purifi 1ET400A EVAL2, implementation shown here. Test conditions were described in post #1. Both amps were loaded with 4 ohm load and driven just close to clipping at maximum music sample amplitude. Those, who checked their sighted preference, have now the amps revealed.

 

[pkane]

In my experience it is always the same case. We are quite sure we can hear a slight difference, which is then not confirmed by the ABX. Then we tend to blame it on the test track or the ABX method. With another test track it is again the same case. OK, we may say that with the test track used or equipment used we could or could not reliably tell the difference.

IME I never found the properly level matched and time aligned test that would reveal sound difference (with music) solely based on harmonic distortion difference. Contrary to that, quite small differences in amplitude frequency response were audible. Which is not the case in this test.

Harmonic distortion levels have to be really high to become audible with music. Easier to detect with test signals.

I didn't pass even the "training" part of the ABX test. What seemed like obvious differences disappeared quickly when switching between A and B even in a sighted test. Sorry, forgot to vote in the poll, but it would've made little difference in the result.

 

[pma]

Thanks for your contribution, Paul. And many thanks for your excellent SW job!!!

 

[MaxwellsEq]

OK, the poll has been closed, with this result:

View attachment 477530

First, thanks to everyone who has contributed to the thread and voted in the poll.

I am missing ABX log reports from those who voted "I can hear a difference". Sadly, such votes without the DBT proof are quite meaningless, at least to my exploration.

Now, the disclosure. Amp X is my class AB A250W amplifier, 2 x 250W/4ohm rated continuous power. Amp Y is Purifi 1ET400A EVAL2, implementation shown here. Test conditions were described in post #1. Both amps were loaded with 4 ohm load and driven just close to clipping at maximum music sample amplitude. Those, who checked their sighted preference, have now the amps revealed.

Thanks for doing the test! I did want to perform the ABX a 2nd time, but got an inner ear infection which would have masked any subtle differences

I replaced an aging Class AB amplifier with a 1ET400 and was unable to hear any difference. The Purifi is an impressive piece of kit in my opinion.

 

[Ashl777]

Maybe a stupid question. If the most sensitive mic currently available is used the measure these amps at listening distance will the wave trace of mic be identical? As the loopback is identical the mic trace should be identical?

 

[antcollinet]

Maybe a stupid question. If the most sensitive mic currently available is used the measure these amps at listening distance will the wave trace of mic be identical? As the loopback is identical the mic trace should be identical?

No - because the recording will be heavily influened by both ambinent noise, and in room reflections. Both will vary from one recording to the next, and the differences from these will far outweigh any differences between amps.