• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Is 22kHz brickwall cut off audible? - Listening test

Can you hear a difference between the files

  • I can hear a difference and I have an ABX result

    Votes: 2 11.1%
  • I can hear a difference but have no ABX result

    Votes: 3 16.7%
  • I cannot hear a difference and I have an ABX result

    Votes: 3 16.7%
  • I cannot hear a difference but have no ABX result

    Votes: 10 55.6%

  • Total voters
    18

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Is 22kHz brickwall cut off audible?

Long term discussion about audibility of >20kHz content, filter slopes, hires, etc. …, so here is my 12 cents. There are 2 test files in 96/24 format, one of them is unmodified, the other one has sharp brickwall cut off at 21.9kHz, but no action below 21.9kHz. The files are located at


To play the files properly, one must be sure that the PC preserves 96kHz playback and does not re-sample down to 44.1kHz or 48kHz due to Windows sound settings. The best way is to use ASIO drivers or to check on DAC display, if it shows the sampling frequency.

If you are interested, please download and unzip the files and try if you can hear the difference and if possible, please also post your ABX test report to verify you were able or not to hear the difference. Please also feel free to use Paul's @pkane excellent Deltawave software that would show you how the files are matched and how they differ.

Following are several Deltawave screenshots of these files:

Original and brickwall spectra

doors_spectra.png


The difference spectrum

doors_spectrumdelta.png


Pkmetric of the test files

doors_pkmetric.png


Report of file matching

DeltaWave v1.0.70, 2022-07-10T22:29:04.5264509+02:00
Reference: BreakOn_1sh.wav[L] 5760000 samples 96000Hz 24bits, stereo, MD5=00
Comparison: BreakOn_3sh.wav[L] 5760000 samples 96000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -160dB
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Hann
Spectrum Window:Kaiser10, Spectrum Size:262144
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Hann, taps:8192, minimum phase=False
Dither:False
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0s, End=0s
Discarding Comparison: Start=0s, End=0s

Initial peak values Reference: -5.115dB Comparison: -5.114dB
Initial RMS values Reference: -24.002dB Comparison: -24.002dB

Null Depth=45.513dB
X-Correlation offset: 1 samples
Drift computation quality, #1: Excellent (0µs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -5.115dB Comparison: -5.116dB
Final RMS values Reference: -24.002dB Comparison: -24.002dB

Gain= 0dB (1x) DC=0 Phase offset=0.006378ms (0.612 samples)
Difference (rms) = -67.88dB [-118.93dBA]
Correlated Null Depth=73.94dB [117.5dBA]
Clock drift: 0 ppm


Files are NOT a bit-perfect match (match=8.28%) at 16 bits
Files are NOT a bit-perfect match (match=0.05%) at 24 bits
Files match @ 49.9967% when reduced to 12.36 bits


---- Phase difference (full bandwidth): 0.0672281438513599°
0-10kHz: 0.00°
0-20kHz: 0.00°
0-24kHz: 0.03°
Timing error (rms jitter): 65.7ns

RMS of the difference of spectra: -149.641964871824dB
DF Metric (step=400ms, overlap=0%):
Median=-50.9dB
Max=-25.9dB Min=-57.7dB

1% > -57.52dB
10% > -54.51dB
25% > -53.44dB
50% > -50.88dB
75% > -47.02dB
90% > -43.26dB
99% > -21.13dB

Linearity 24.5bits @ 0.5dB error
PK Metric (step=400ms, overlap=50%):
RMS=-116.6dBr
Median=-130.0
Max=-89.0

99%: -111.84
75%: -128.55
50%: -130.04
25%: -131.55
1%: -134.86

gn=0.999999396412136, dc=3.00935242754951E-09, dr=0, of=0.6122974419

DONE!

Signature: ab5dba637a1bce48d8a161f522e87103

And finally my ABX test report

Code:
foo_abx 2.0.2 report
foobar2000 v1.4.8
2022-07-10 22:17:35

File A: BreakOn_1sh.wav
SHA1: 73fd67f8fb135adf5e0ec92bc9f2178bbe269709
File B: BreakOn_3sh.wav
SHA1: 80aebddcdc26e9851a9441ddc89aae59a2d099eb

Output:
ASIO : Topping USB Audio Device
Crossfading: NO

22:17:35 : Test started.
22:19:16 : 01/01
22:19:32 : 02/02
22:19:49 : 03/03
22:20:23 : 04/04
22:20:41 : 05/05
22:20:58 : 06/06
22:21:14 : 07/07
22:21:32 : 07/08
22:21:54 : 07/09
22:22:15 : 08/10
22:22:15 : Test finished.

 ----------
Total: 8/10
Probability that you were guessing: 5.5%

 -- signature --
1fecd71c51663eee539fc9f3fb52be641d5de76f

Code:
foo_abx 2.0.2 report
foobar2000 v1.4.8
2022-07-11 10:03:01

File A: BreakOn_1sh.wav
SHA1: 73fd67f8fb135adf5e0ec92bc9f2178bbe269709
File B: BreakOn_3sh.wav
SHA1: 80aebddcdc26e9851a9441ddc89aae59a2d099eb

Output:
ASIO : Topping USB Audio Device
Crossfading: NO

10:03:01 : Test started.
10:05:08 : 01/01
10:05:28 : 02/02
10:05:43 : 02/03
10:06:09 : 03/04
10:06:23 : 04/05
10:06:39 : 04/06
10:06:53 : 05/07
10:07:13 : 06/08
10:07:27 : 07/09
10:08:06 : 08/10
10:08:06 : Test finished.

 ----------
Total: 8/10
Probability that you were guessing: 5.5%

 -- signature --
aadf51dde71992c4a987ec42a7e1476dfe2ddba2

16/20 total result :)

Have fun!
 
Last edited:

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
3,687
Likes
4,068
Is 22kHz brickwall cut off audible?

Long term discussion about audibility of >20kHz content, filter slopes, hires, etc. …, so here is my 12 cents. There are 2 test files in 96/24 format, one of them is unmodified, the other one has sharp brickwall cut off at 21.9kHz, but no action below 21.9kHz. The files are located at


To play the files properly, one must be sure that the PC preserves 96kHz playback and does not re-sample down to 44.1kHz or 48kHz due to Windows sound settings. The best way is to use ASIO drivers or to check on DAC display, if it shows the sampling frequency.

If you are interested, please download and unzip the files and try if you can hear the difference and if possible, please also post your ABX test report to verify you were able or not to hear the difference. Please also feel free to use Paul's @pkane excellent Deltawave software that would show you how the files are matched and how they differ.

Following are several Deltawave screenshots of these files:

Original and brickwall spectra

View attachment 217662

The difference spectrum

View attachment 217663

Pkmetric of the test files

View attachment 217664

Report of file matching



And finally my ABX test report

Code:
foo_abx 2.0.2 report
foobar2000 v1.4.8
2022-07-10 22:17:35

File A: BreakOn_1sh.wav
SHA1: 73fd67f8fb135adf5e0ec92bc9f2178bbe269709
File B: BreakOn_3sh.wav
SHA1: 80aebddcdc26e9851a9441ddc89aae59a2d099eb

Output:
ASIO : Topping USB Audio Device
Crossfading: NO

22:17:35 : Test started.
22:19:16 : 01/01
22:19:32 : 02/02
22:19:49 : 03/03
22:20:23 : 04/04
22:20:41 : 05/05
22:20:58 : 06/06
22:21:14 : 07/07
22:21:32 : 07/08
22:21:54 : 07/09
22:22:15 : 08/10
22:22:15 : Test finished.

 ----------
Total: 8/10
Probability that you were guessing: 5.5%

 -- signature --
1fecd71c51663eee539fc9f3fb52be641d5de76f

Code:
foo_abx 2.0.2 report
foobar2000 v1.4.8
2022-07-11 10:03:01

File A: BreakOn_1sh.wav
SHA1: 73fd67f8fb135adf5e0ec92bc9f2178bbe269709
File B: BreakOn_3sh.wav
SHA1: 80aebddcdc26e9851a9441ddc89aae59a2d099eb

Output:
ASIO : Topping USB Audio Device
Crossfading: NO

10:03:01 : Test started.
10:05:08 : 01/01
10:05:28 : 02/02
10:05:43 : 02/03
10:06:09 : 03/04
10:06:23 : 04/05
10:06:39 : 04/06
10:06:53 : 05/07
10:07:13 : 06/08
10:07:27 : 07/09
10:08:06 : 08/10
10:08:06 : Test finished.

 ----------
Total: 8/10
Probability that you were guessing: 5.5%

 -- signature --
aadf51dde71992c4a987ec42a7e1476dfe2ddba2

16/20 total result :)

Have fun!
Very strange, but well done!
Does one of the files sound better to you, or just different?
 
  • Like
Reactions: pma
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Very strange, but well done!
Does one of the files sound better to you, or just different?
I agree that it is strange ;). And the files are strange, aren't they. Vinyl rip with all technical imperfections, only my phono preamp and ADC (E1DA) are perfect. My long term experience - in cases like this the "perfect" sound recording may not be most revealing regarding sound differences.

To answer your question - the files sound slightly different to me (I use 20s - 40s time interval to test). The original file sounds slightly more sharp, more aggressive to me, the cut file slightly mellower. If I listen to the difference file (produced by Deltawave) only, I hear nothing. If the difference file is slowed down to 48kHz, then it is possible to hear that the remaining difference is in transients, whatever is their origin (vinyl track, cartridge resonance ...., who knows). But cut it and there may be a difference. Ears, headphones, whatever ... ? :D
 

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
I agree that it is strange ;). And the files are strange, aren't they. Vinyl rip with all technical imperfections, only my phono preamp and ADC (E1DA) are perfect. My long term experience - in cases like this the "perfect" sound recording may not be most revealing regarding sound differences.

To answer your question - the files sound slightly different to me (I use 20s - 40s time interval to test). The original file sounds slightly more sharp, more aggressive to me, the cut file slightly mellower. If I listen to the difference file (produced by Deltawave) only, I hear nothing. If the difference file is slowed down to 48kHz, then it is possible to hear that the remaining difference is in transients, whatever is their origin (vinyl track, cartridge resonance ...., who knows). But cut it and there may be a difference. Ears, headphones, whatever ... ? :D
So you do a brick wall by FFT filtering? Thats not the typical way.... (And could explain audibility)
 
Last edited:

Jimbob54

Grand Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
11,063
Likes
14,694
Given the current hot topic on ASR is the Chord Mscaler with its (impressive) v sharp/ brickwall this is interesting. But also worth pointing out that the Mscaler is meant to be used in conjunction with Chord DACs that already implement a pretty steep and deep filter.

I can credit being able to detect a 22k brick wall vs very slow or no filter, less believing of claims that people can differentiate a very steep vs an even steeper good filter.
 

Andysu

Major Contributor
Joined
Dec 7, 2019
Messages
2,883
Likes
1,477
above 20KHz didn't think anyThing was there expect what caters for dogs cats and pet bats .
life above 20KHz not as we know it , jim ?


maybe i play a bit of lp and cd later and see what only frequency range on RTA will allow
 
Last edited:

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
I just did a diff between the two tracks: clearly audible difference in the full audio band...(even distorted) There must be something wrong with your way of filtering..

The diff:
1657551890757.png
 

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,312
Likes
4,425
Location
Suffolk Coastal, UK
My ears are shot to hell and the aids not 'resolving enough' up top. To me using HD265's which fit round my ears not on them, the distortion on Jim's vocals is terrible and to me, unlistenable I'm afraid (I'm a touch too young to 'get' The Doors as a few pals older than me did... Both tracks sounded awful to me, so best leave it there, sorry... In any case, 48/24 is the best this thing here can do, so moot...
 

tmtomh

Major Contributor
Forum Donor
Joined
Aug 14, 2018
Messages
2,635
Likes
7,486
So you do a brick wall by FFT filtering? Thats not the typical way.... (And could explain audibility)

I think this question - is the filter implemented here in a typical way - needs to be answered clearly and definitively before it's worth even engaging in this test.
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
2,916
Likes
3,831
I'm skeptical and I'd chalk-up any audible difference to a side-effect of the filtering. But, I didn't bother with the tests because I'm over 60 years and although I don't know exactly where my hearing cuts-off, I can't hear to 20kHz. (Plus, there are several psychoacoustic reasons why we can't normally hear the highest harmonics in music.)

The original file sounds slightly more sharp, more aggressive to me, the cut file slightly mellower.
It would be better if you can use real words like noise, distortion, or frequency response. ;) Or, you could say more bass, or less highs, etc. You can't measure or define +3dB of "sharpness".

I just did a diff between the two tracks: clearly audible difference in the full audio band...(even distorted) There must be something wrong with your way of filtering..
If you are doing simple subtraction that's not reliable...

If you get silence that does prove there is no difference. If you don't get silence there is a difference in the file but not necessarily a difference in sound. The most obvious example is if you add one millisecond to the beginning that doesn't change the sound but it makes a "loud" comb filtered difference file. Filtering can cause phase-shifts in the pass band with a somewhat similar (but usually less drastic) result. If you subtract an MP3 from the original the result is also "much worse" than the difference in the sound. (MP3 also adds a delay but that part can be compensated for.)

In other words, "The sound OF the difference isn't the same as the difference IN the sound." ;)
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
I don't know how you did the filtering, but I just resampled file #1 to 48KHz, then back to 96KHz in Audacity (triangle dither). Quick and easy to replicate. When I did this I only hear light hiss when playing the delta in DeltaWave at +60dB boost (whereas when playing the delta of the other files I heard music) . The PK metric is 122dB.

So while I can't explain why the other filtering gives an audible difference, I think it can be inaudible.
1657555075008.png
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I just did a diff between the two tracks: clearly audible difference in the full audio band...(even distorted) There must be something wrong with your way of filtering..

The diff:
View attachment 217732

Perhaps the 2 files are not time aligned (which Audacity does not do but Paul's software can) which could account for the nulling error.

That said, the sound quality of the recording is terrible. Only the needle drop and first 10 seconds sound somewhat pleasing.
I am distracted/annoyed way too much by the distortion and can't hear above 15kHz anyway.

I would say... use a well recorded 96kHz recording and redo the test ?
 

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
would be better if you can use real words like noise, distortion, or frequency response. ;) Or, you could say more bass, or less highs, etc. You can't measure or define +3dB of "sharpness".

If you are doing simple subtraction that's not reliable...

If you get silence that does prove there is no difference. If you don't get silence there is a difference in the file but not necessarily a difference in sound. The most obvious example is if you add one millisecond to the beginning that doesn't change the sound but it makes a "loud" comb filtered difference file.
I checked the time alignement, it was of by one sample and I corrected for that before subtracting
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
I checked the time alignement, it was of by one sample and I corrected for that before subtracting

Sample alignment is insufficient for a proper difference file. The error in the result can be very large if you don't align to a tiny fraction of a sample.
 

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
Sample alignment is insufficient for a proper difference file. The error in the result can be very large if you don't align to a tiny fraction of a sample.
sub sample timing errors will only give a real difference at high frequencies.. What I heard was a difference with the full spectrum. maybe there is also a gain error?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Sample alignment is insufficient for a proper difference file. The error in the result can be very large if you don't align to a tiny fraction of a sample.
Question: Is this also the case when a single digital file is used ?
This sure will be the case when 2 recorded files are nulled from separate recordings (the clocks will drift).
In this case only one file is recorded and the second one is the same file but filtered there should be no clock drift.
 
Top Bottom