Tracking my listening tests

[Pdxwayne]

I am using this to track my listening test results as posted in other threads.

Online blind tests using my Samsung Note 9 with headphone.
0.5db - pass
0.2db - can't do up
5c pitch - pass
2 ways, 10ms - pass
Screenshots in https://www.audiosciencereview.com/...ds/how-good-is-your-hearing.20417/post-692530

online blind test using Topping e30, Parasound A21, Paradigm Persona B
0.5db - pass
Screenshot in
https://www.audiosciencereview.com/...my-gustard-x16-measurements.20679/post-690212

ABX - compare orig with orig eq with slope of ~0.7db from ~1.3 kHz to ~10khz
Topping E30, Topping L30, AKG 371
13/16
Screenshots at
https://www.audiosciencereview.com/...no-v-stereo-what-is-audible.21864/post-821207

ABX - compare orig with orig eq down ～0.3db from ~1.3 kHz to ~10khz
Topping E30, Topping L30, minidsp 2x4, Parasound A23, Pioneer bookshelf
12/16
Screenshots
https://www.audiosciencereview.com/...no-v-stereo-what-is-audible.21864/post-821300

ABX - compare orig with orig adjust right channel down 0.2db
Topping E30, Topping L30, direct to Parasound A23, Pioneer bookshelf
13/16
Screenshots
https://www.audiosciencereview.com/...my-gustard-x16-measurements.20679/post-821965

 

[Pdxwayne]

My first double blinds DAC comparison was between KTB and E30. I learned a lot. I could not tell them apart when they were voltage matched up to ~5khz. My initial results are at https://www.audiosciencereview.com/...topping-e30-listening-tests.17988/post-619037. I am now thinking I might not have the proper songs for testing high frequencies, thus failed to tell them apart. Not going to go back to do another blind test to check that.
: P

 

[Pdxwayne]

My second double blinds DAC test was between Gustard x16 and E30. Learned more again.
Only did 6 total tests and gave up.
Got 5 out of 6. See https://www.audiosciencereview.com/...my-gustard-x16-measurements.20679/post-699296

As mentioned in my last ABX result in post #1, I got 13/16 (p-value ~1%) in my subsequent ABX tests in https://www.audiosciencereview.com/...my-gustard-x16-measurements.20679/post-821965. Combined both results, I would say indeed I can sense 0.2db channel imbalance.

 

[Pdxwayne]

This is not about my tests. But I want to keep the link of hearing threshold thread handy here:
https://www.audiosciencereview.com/...-of-amp-and-dac-measurements.5734/post-127757

 

[Pdxwayne]

Here is one interesting ABX.

A song is recorded using DAC and ADC chain that has THD+Noise of -104db.
Can one tell that song apart from original file, when played back in a chain with DAC+Amp with SINAD of about 112db?

This is the THD+Noise of the recording chain:

ABX the orginal file vs captured file, playing back via USB to E30 to L30 to AKG K371. Able to get 8 out of 8.

BTW, the DAC that did the capture is Topping D30pro. The ADC is Focusrite Forte.

Per https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-849648, the PKmetric of d30pro vs original is at -68db. Based on my results, PKmetric of -68db means that difference is audibled for some people.

 

[solderdude]

Did you use the non USB D30 pro recording ?
Are the outputs level matched ?

Attenuating music by -68dB usually results in complete silence so when differences are this small and not caused by weird additions in music (listen to the null) then -68dB difference cannot be audible. -68dB IM products ... sure these can be audible.

 

[Pdxwayne]

Did you use the non USB D30 pro recording ?
Are the outputs level matched ?

Attenuating music by -68dB usually results in complete silence so when differences are this small and not caused by weird additions in music (listen to the null) then -68dB difference cannot be audible. -68dB IM products ... sure these can be audible.

Yes, if you see the title, it is from the optical capture. If you see the abx screen capture, you can see that ABX has a feature to auto level match.

Note that -68db does not mean that the whole song difference is flat at -68db, see that chart in https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-849648 again please? There are section where it was like -61db.

 

[solderdude]

The point is that the difference may well be -61dB in level between them but if the null sounds the same (but louder) then the wave form is very closely the same.
I will try to explain.
60dB is a factor 1000 smaller.
Assume peaks you measured were 2V. A factor 1000 lower = 2mV.

When I null and there is a difference of -60dB as the result you will get that same result when one file = 2V peak and the other file is 1.998V.
So... this is 0.0087 dB level difference.
You established 0.2dB as your detection limit.
Can you detect 0.009dB as well ?
This is what I mean with understanding what nulling means and what the plots tell you.
That's also why I urged you to listen to the null. When that doesn't sound ' musical' the differences are merely level or phase differences resulting in minuscule level differences that are impossible to detect.

When nulling and the null sounds nasty and is -60dB down then the distortion products that are in that null and have no relation to the original sound then yes, they could be audible.
When the differences are only small level differences exist -30dB is not even audible.
Even automatic detection can be 0.1% off in amplitude depending on how/where peak detection takes place and 2 files are normalized to the same value

That's what is meant by: be careful with interpretation of null results.

 

[dominikz]

Here is one interesting ABX.

A song is recorded using DAC and ADC chain that has THD+Noise of -104db.
Can one tell that song apart from original file, when played back in a chain with DAC+Amp with SINAD of about 112db?

This is the THD+Noise of the recording chain:
View attachment 141879

ABX the orginal file vs captured file, playing back via USB to E30 to L30 to AKG K371. Able to get 8 out of 8.
View attachment 141882

BTW, the DAC that did the capture is Topping D30pro. The ADC is Focusrite Forte.

Per https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-849648, the PKmetric of d30pro vs original is at -68db. Based on my results, PKmetric of -68db means that difference is audibled for some people.

Hello, an interesting comparison indeed - I have to applaud the determination! Though I agree with others that it will be difficult to draw conclusions without more work, and even then the results might be inconclusive.

With regard to the ABX - I see from the foobar ABX comparator log that the automatic gain adjustment was different for both files (+5,5dB on one and +3.11dB on the other) - implying that the initial loudness was not equal when recording the files.
Since we don't know which algorithm the ABX comparator plugin uses to level-match, may I suggest to instead try and level-match the DAC outputs playing a sine tone with a true RMS multimeter before recording the files in the first place? Then you wouldn't need to use the gain alignment in foobar ABX comparator plugin.
If you don't have a true RMS multimeter you could also use the Focusrite Forte and REW (RTA view) to level match - since you're using the same ADC in both cases it should work OK.

IMHO it might also theoretically help with null-test results (though perhaps you've already taken that into account - I haven't read the whole thread).

Good luck and have fun

EDIT: Sorry, I was brought here by the link from the other thread and notice now my response may have been better suited there.

 

[Pdxwayne]

Hello, an interesting comparison indeed - I have to applaud the determination! Though I agree with others that it will be difficult to draw conclusions without more work, and even then the results might be inconclusive.

With regard to the ABX - I see from the foobar ABX comparator log that the automatic gain adjustment was different for both files (+5,5dB on one and +3.11dB on the other) - implying that the initial loudness was not equal when recording the files.
Since we don't know which algorithm the ABX comparator plugin uses to level-match, may I suggest to instead try and level-match the DAC outputs playing a sine tone with a true RMS multimeter before recording the files in the first place? Then you wouldn't need to use the gain alignment in foobar ABX comparator plugin.
If you don't have a true RMS multimeter you could also use the Focusrite Forte and REW (RTA view) to level match - since you're using the same ADC in both cases it should work OK.

IMHO it might also theoretically help with null-test results (though perhaps you've already taken that into account - I haven't read the whole thread).

Good luck and have fun

EDIT: Sorry, I was brought here by the link from the other thread and notice now my response may have been better suited there.

Both x16 and d30pro can't be level matched easily. They can be adjusted only in increment of 0.5db, but their difference is between 0 and 0.5db. Also, both has different level of channel imbalance. One has imbalance of 0.06db, the other has ~0.2db. Lastly, in order to make sure ADC not distort due to peak, when recording, I adjusted both DAC volume to 5db down. Thus, you see my capture files has lower dB than the original file.

Fortunately Delaware can level match each channel of comparison before performing null calculation. Good thing foobar2000 can also level match before ABX.

 

[dominikz]

Both x16 and d30pro can't be level matched easily. They can be adjusted only in increment of 0.5db, but their difference is between 0 and 0.5db. Also, both has different level of channel imbalance. One has imbalance of 0.06db, the other has ~0.2db.

Ah, I see, that's unfortunate
You might be able to get a better input level match by running both at the same output dBFS level and then using an external passive attenuator to match levels - though I understand this is not something everyone has laying around at home.

Matching input levels as closely as possible before recording would ensure slightly better control of variables. However it can also introduce some additional uncertainty - e.g. one would have to check that introducing an additional device (attenuator) in the signal flow didn't affect one DUT more than the other; due to impedance mismatch or similar.

It can be very difficult to control all parameters (especially without specialized HW/SW) - but measurements and test protocols can still be quite fun and informative to do. Though it is somewhat of a bottomless pit, depending how deep you want to dive into it

 

[Pdxwayne]

A good site to calculate probability of ABX success by chance:
https://stattrek.com/online-calculator/binomial.aspx

 

[Pdxwayne]

I have done one interesting "impossible" ABX. It is very difficult. Only got 30/48. Still, that result points to about 5% chance of guessing, better than all the participants there so far, who got no better than pure guessing results?

The screenshots and the test thread link at

https://www.audiosciencereview.com/...-preamplifier-about-2-7-thd.18392/post-856169

 

[Pdxwayne]

ABX for 2 amps. I was able to sense the small channels imbalance more accurately halfway through the ABX. Got 25/32.

Per OP's response, the right channel difference is 0.3db. Left is 0.1db. My results showed that I can sense single channel imbalance of 0.3db with real music.

See https://www.audiosciencereview.com/...files-recorded-for-download.25646/post-876181

 

[Pdxwayne]

Another ABX.

Audibility of harmonic distortion of the pure sine wave that went through a tube preamplifier, resulting in THD = 0.85%

8 out of 8

https://www.audiosciencereview.com/...0-85-from-tube-preamplifier.26299/post-900833

 

[Pdxwayne]

Another ABX.

Can I sense a slight difference in dB in ~30hz sub bass between right and left channel?

One member was kind enough to provide files to ABX. One file original and the other left and right channel flipped.

Results is 14/16.

Link:

Stereo Bass using subwoofers

OK, I am almost finished with my real work today, so I will do it and put them up in a little while. Mind if I cut the track down to the first 1.5min? Actually, it is best to cut the first minute out, then use just the second minute. One minute clips from 60 seconds to 120 seconds should be...

www.audiosciencereview.com

 

[Pdxwayne]

ABX of stereo subs using files with different loudness between left and right.

Files 9 and 10: 20 dB of L/R difference applied
Files 11 and 12: 9 dB of L/R difference applied
Files 13 and 14: 3dB of L/R difference applied
Files download link: https://drive.google.com/drive/folders/1t2wkzjK0GNtX4NOadek45gKRJPVE7xqr

Able to sense directional sub tones from left vs right for all 3 sets. All 10/10 results.

Link:

Stereo Bass using subwoofers

Hmm, this particular one is about difference in 32 to 34 Hz. If I turn off ARC, then right subs will have significant dB peak due to room mode, and left sub will have significant null due to room mode. The difference, based on the measurements, should be in the +10db range. This +10db imbalance...

www.audiosciencereview.com

 

[Pdxwayne]

Further stereo subwoofers sub bass directions ABX.

This time made sure harmonics above 80hz are too low to give any clues of directions.

Although much harder, still got 10/10.

Link:

Stereo Bass using subwoofers

May I know in any of your prior tests, did it involves artificial low sub bass in the 30hz? See the description of the files: https://www.audiosciencereview.com/forum/index.php?threads/stereo-bass-using-subwoofers.11034/post-928692 The files are all filter at 48db sloop at 80hz. The main...

www.audiosciencereview.com

 

[Pdxwayne]

Vinyl rip vs CD ABX. Headphone setup 10/10. Speakers setup 18/20.

Link

One more Vinyl rip x digital data ABX listening test

The test files are here: http://pmacura.cz/alaba.zip It is a vinyl rip file and the same master in digital data. Time and level aligned as much as possible. If you are interested, please listen to the files and post your ABX result and try to play fair, it means please listen under normal...

www.audiosciencereview.com

 

[Pdxwayne]

Simple 2 tones listening tests. 250hz and 1250hz at the same time. 1250hz is at -60db. Most people should be able to barely sense the difference.

Topping E30 to Topping L30 to AKG 371. 8/8.
Gustard x16 to Gustard H16 to AKG 371.
8/8.

Link for Gustard abx results:

Can you hear a difference? 250hz and 1250hz playing at same time, but 1250 Hz is at 60db down. Vs pure 250hz tone.

Even easier on the pixel 5 internal speaker using a free abx app I found. Can't miss the high tone on that. Can you try using a decent headphones? Thanks!

www.audiosciencereview.com