@MarcelvdG - I am generating the waveform by summing 2,000 harmonics of 10Hz. Each harmonic's amplitude and phase is adjusted individually. The results exactly correspond to +3dB at 200KHz.
Ed
Ed
The method I use in my spreadsheet leads to 20 dB • 10log(√2) ~= +3.010299... dB at 200 kHz and it calculates only one filtered impulse rather than an infinite series of them. It's not exactly the same, but it seems pretty close.
The impressions on audibility are quite pointless without having a statistically significant result of the double blind test. Has anyone posted such report?
You are right PMA. But I'm not holding my breathe waiting for someone to post such a report. If no one had claimed they could 'hear' EdGr's item 4, I would have said it was 'proven' and no need for DBLTs.
But as there are people who claimed they can, I'm now racking my brains to dream up a way to DB test myself with headphones. That's cos even the best ears (and I've had some of the best in my DBLT panel) are as opinionated and prejudiced as the most rabid wannabe Golden Pinnae. I trust their opinions blind but not always sighted.
Several things svp described, eg headphones and a particular volume level for max discrimination, I also encountered in my tests on somewhat similar phenomena .. so I'm open to svp actually being able to hear the 'difference'.
It's perhaps terribly crude, but ...
If you can split the provided file into the segments necessary, you could perhaps play those "on random" through your chosen audio player.
Note your findings.
Go back through the play history to see the order of playback.
Did test again. Conditions the same as in post #34 https://www.diyaudio.com/community/threads/take-the-click-and-pop-challenge.424691/post-7949926
Start to hear "4" on ~26/100 volume, clearly on 30/100 and more. As I know myself in the evening my hearing becomes less sensitive, so probably hearing threshold will be 35/100 of PC volume.
That is still very loud for me to listen, even 25/100 is much louder I typically listen.
Start to hear "4" on ~26/100 volume, clearly on 30/100 and more. As I know myself in the evening my hearing becomes less sensitive, so probably hearing threshold will be 35/100 of PC volume.
That is still very loud for me to listen, even 25/100 is much louder I typically listen.
Ed, frankly, I am confused what is asked to be done. I only take into account results of ABX tests. So I have split your single test file into 4 files of same length, and denoted them test1 - test4 according your description above. I have uploaded the files to
https://pmacura.cz/testclicks.zip
Now I would suppose to compare one file against another file. I did ABX of (2) against (4), test2 against test4. The result is below.
Code:
foo_abx 2.1 report
foobar2000 v2.0
2025-03-07 15:04:37
File A: test2.wav
SHA1: 18b3f71b7d69d67ec26c7e9104d460360080af25
File B: test4.wav
SHA1: 077344e1d886d2a88202f1f9eed6483cbb460f1e
Output:
ASIO : Topping USB Audio Device
Crossfading: NO
15:04:37 : Test started.
15:05:34 : 01/01
15:05:39 : 02/02
15:05:46 : 03/03
15:05:53 : 04/04
15:05:58 : 05/05
15:06:04 : 06/06
15:06:09 : 07/07
15:06:14 : 08/08
15:06:22 : 09/09
15:06:27 : 10/10
15:06:32 : 11/11
15:06:41 : 12/12
15:06:46 : 13/13
15:06:51 : 14/14
15:06:56 : 15/15
15:07:01 : 16/16
15:07:01 : Test finished.
----------
Total: 16/16
p-value: 0 (0%)
-- signature --
bb6c131e89f6cc45ce7716133d506a7012692b8eIs it what you asked for? Or should it be another 2 files?
It is easy to hear 2kHz below the impulses, in periodic time intervals. However, to stay in the field of evidence, we should be supporting our hearing impressions by abx report facts.
So yes, there is a difference. But I am still lost in conclusions that could be done from this fact. I can see no consequences with MM preamplifiers design topologies, when taking into account real cartridge output and not the artificial signal games.
Well, if someone makes an automated test of "4" vs "4 without the beep" - I would like to participate. But there is no need for that as I can hear it clearly. Maybe if I do test like 20000x in a row I get bored or not attentive enough, or my hearing gets numb by the test itself and I will make some mistake, making it a 99.9% correct, not 100%.
Also another factor is not discussed, I think: we already know what to listen for. If the task was to hear some common sound, like different bird songs, car roars, other well known sounds in the background and then ask what you heard in the background - the task gets much more complex.
More complex also are the 320kbits mp3 tests vs CD quality - in most tracks I hear no difference. But now I have couple of good DACs and probably will search internet to check again. Headphones will be the same Superlux.
I would like to get another file of "4" with -3dB or -6dB level from this signal.
I agree, it is very very audible. No need to extract the beeps. I thought that the beeps in (4) should have been the difference from the (2) file, that is why did ABX (2) x (4). The audible difference was just those periodic beeps. But this is irrelevant.
My question is different, what is supposed to be tested???
Anything like this can be easily done, but why??
@PMA - A blind test requires more effort. I want to keep the test short enough that people will try it.
(2) tests the ability to hear the +1 error.
(4) tests the ability to hear a signal at 0.01% (-80dB). The 2KHz sine is a proxy for any low-level signals.
You guys are doing really well. I could not hear either on my main system.
Ed
(2) tests the ability to hear the +1 error.
(4) tests the ability to hear a signal at 0.01% (-80dB). The 2KHz sine is a proxy for any low-level signals.
You guys are doing really well. I could not hear either on my main system.
Ed
I assume I would not hear it on my main system with speakers, too. Ambient noise.
I used JBL460 headphones, Topping D10s DAC and my design headphone amplifier
https://www.audiosciencereview.com/...-amplifier-distortion-test-files-added.21014/
It is really quite audible with headphones.
Last edited:
If you could provide an extra file with just the clicks, so no zero and no tone, then people who are prepared to put in the extra effort could do a blind test with an ABX tool. They can then just test section 2 or 4 from the file in post #1 against the file with only clicks.
Here is a file of only the clicks for people who want to do ABX against the original file.
Everyone else - don't use this file.
Ed
Thank you for the file. Below is the result of part (4) of the original file against the clicks only. 16/16 correct answers. The difference is in the repeated beeps, though they are very silent.
Code:
foo_abx 2.1 report
foobar2000 v2.0
2025-03-07 21:54:15
File A: test.flac
SHA1: c890c0cf2b8b42bf7dc077437edef17af09a949f
File B: clickonly.flac
SHA1: 9f9fcfbecfceca9527b9bf96b5ac9db288fcb1b0
Output:
ASIO : Topping USB Audio Device
Crossfading: NO
21:54:15 : Test started.
21:55:58 : 01/01
21:56:07 : 02/02
21:56:16 : 03/03
21:56:23 : 04/04
21:56:32 : 05/05
21:56:40 : 06/06
21:56:47 : 07/07
21:56:54 : 08/08
21:57:01 : 09/09
21:57:07 : 10/10
21:57:15 : 11/11
21:57:22 : 12/12
21:57:29 : 13/13
21:57:36 : 14/14
21:57:43 : 15/15
21:57:50 : 16/16
21:57:50 : Test finished.
----------
Total: 16/16
p-value: 0 (0%)
-- signature --
7789fccbcb564b9a3e0eaee5997c79d16ceb59efNote: I did not vote in the poll.
Last edited: