[attachment=8072:Meridian...eaker_FR.png]
[attachment=8071:PC351_wi...0SC00-04.png]
[attachment=8072:Meridian...eaker_FR.png]
[attachment=8071:PC351_wi...0SC00-04.png]
[attachment=8073:audibili...al_fig_2.png]
Frequency Response
[attachment=8103:FR_DA_FS.gif]
Dynamic Range
[attachment=8099:DR_DA_FS.gif]
Jitter
[attachment=8101:JIT_DA_FS.gif]
S/N Ratio
[attachment=8102:SNR_DA_FS.gif]
[attachment=8164:Stereoph...ts_cover.png]
That is truly hilarious. Where did you get that?
That is truly hilarious. Where did you get that?
I cannot tell a lie - I found the original in the pictures folder of this PC. ;-)
[attachment=8218:swisha1644.wav]
This is an interesting little file of a signal that is called "A Swish" that just about any FFT analysis will portray as having a flat PSD 20-20 KHz.
It largely avoids the bucket alignment problem with mulitones, and the integration time problem with pseudorandom signals.
If one needs a wide bandwidth version of this, just change the .wav file header to a higher sample rate like 96 KHz.
And while I'm at it - an attempt at a bucket-aligned multitone
[attachment=8219:bucket_a...one_2496.wav]
Here are two approx 10 second samples of 19 KHz and 19_20 KHz test tones re-recorded by looping through a M-Audio 25192 audio interface running under 64 bit Win 7.1 with all current windows updates.
19 KHz test tone
[attachment=8220:19_KHz_o..._ap24192.flac]
19+20 KHz test tones
[attachment=8221:19_20_KH...192_2496.flac]
Looks like multiple tones are a more difficult test
and more:
2496 swisha:
[attachment=8227:swisha_2496.flac]
2496 swisha looped through AP24192
[attachment=8226:swisha24..._ap24192.flac]
Bucket aligned multitone looped through AP24192:
[attachment=8225:bucket_a...192_2496.flac]
CEP 44.1 KHz transition bands (Quality=0 to Quality=100):
[attachment=8231:CEP_44.1...n__bands.png]
Details enlarged:
[attachment=8230:CEP_44.1...enlarged.png]
Mislabeling alert - the resampling quality in the quality = 100 files was actually quality = 999
The following shows a comparison of the impulse response of the two files
[attachment=8232:impulse_...it_1-999.png]
More detail:
[attachment=8234:impulse_...9_expand.png]
This is the data file:
[attachment=8233:impulse_...it_1-999.flac]
Comparison of Sox and CEP 2.1 resampling of 24192 impulse:
Sox resampling:
[attachment=8238:sox_resa...r_wombat.png]
Audacity resampling:
[attachment=8240:Audacity...sampling.gif]
Ringing anybody? ;-)
CEP 2.1 resampling Q=30:
[attachment=8239:CEP_resa..._per_ABK.png]
CEP 4416 Quality=100 tests
Transition band:
[attachment=8245:24192tes...00_24192.png]
Impulse response:
[attachment=8246:24192tes...00_24192.png]
Latest keys jangling 2496 tests with IM test tones starting at 12 seconds that are only needed for testing your monitoring system for IM, not needed for the listening test of downsampling.
2496 Reference file:
[attachment=8248:keys_jan...tones_f4.flac]
The above file downsampled to 1644 and upsampled to 2496:
[attachment=8247:keys_jan...tones_f4.flac]
For those who are not familiar with these files, the first 12.5 seconds is the keys jangling test sound, designed to elicit audible differences due to 16/44 versus 24/96.
Do not apply Replaygain. normalizing or other leveling techniques to these files. Their content 20-17 KHz is correctly level matched.
Beware! These files can fool you. They contain Full Scale test signals that are either not very loud or inaudible. You could conceivably turn your monitoring system up too far and hurt something. They should cause no problems if played at reasonable listening levels. OTOH, some people can't hear 95% or more of these files due to high frequency hearing loss. Hearing nothing, they could conceivably turn their systems up past the breaking point. Don't be that guy!
The remainder of the files are designed to elicit audible misbehavior from monitoring systems that could falsify the key jangling test.
They are composed of a highly audible frequency test tone that you are supposed to hear, followed by 3 sets of ultrasonic twin tones that will elicit no audible response from a monitoring system with adequate quality. They are separated by 2 tics. If your monitoring system has high frequency IM that will invalidate any results from the keys jangling comparisons, you will not hear silence between the tics. Exactly what you will hear is unpredictable, but field reports include hearing low tones similar to the first audible tone, to all sorts of garbagy and distorted sounds. If your monitoring system is adequate you should be unable to hear any difference between the two test files when playing these ultrasonic tones. Both files will produce pure silence between the tics.
By the way, it seems like a lot of people don't seem to want to take advantage of the following:
[attachment=8249:JAES_Cla...Criteria.png]
Which is not presented ideally and therefore hard to interpret. It is based on listening tests both sighted and blind that were done to determine how well the FR of UUTs have to be matched so that the FR is not a reasonable explanation for any audible differences that are found. The commonly heard +/- 0.1 dB 20-20 KHz tolerance that is often given is overly-strict. Better to err on the safe side, I guess.
At any rate the way to interpret this chart for the purpose at hand is to understand that there is a built-in safety factor. We're not saying that FR differences outside the limits show on the chart are for sure audible all the time, we're saying that if you are inside these limits, you are safe.
The curves themselves may not be straight forward. The basic idea is the observation that FR differences over narrow bands are far less audible than FR differences over wide bands. Furthermore FR differences at the frequency extremes are far less audible than those near the midband.
For example the chart says that a >5 dB difference over 1/3 octave centered at 20 KHz is safe. That means that a FR error would have to be several times larger than this to be audible. At least twice.
So that is license to let the FR drop by 10 dB or more at 20 KHz if over only a 1/3 octave band. 1/3 octave centered at 20 KHz is 6 KHz wide at the -3 dB points. Allowing response to drop by something like 10 dB @ 17 KHz and 6 dB or more @ 20 KHz can thus be reasonably expected to be inaudible.
CEP 4416 Quality=100 tests
Transition band:
[attachment=8245:24192tes...00_24192.png]
Impulse response:
[attachment=8246:24192tes...00_24192.png]
If you want to use a picture of an impulse response as a means to estimate how audible ringing
might be, then you should plot power (in dB) vs. time; e.g. see the graphs at the bottom of the page here: http://sox.sourceforge.net/SoX/Resampling (http://sox.sourceforge.net/SoX/Resampling)
If you want to use a picture of an impulse response as a means to estimate how audible ringing might be, then you should plot power (in dB) vs. time; e.g. see the graphs at the bottom of the page here: http://sox.sourceforge.net/SoX/Resampling (http://sox.sourceforge.net/SoX/Resampling)
The chart at the top of that page illustrates the conceptual problem. It portrays signals that are 90 dB down and are also > 20 KHz as if they are easily perceptible by means of making them as visible as signals at FS.
Arnold, I tested the keys_jangling_full_band_2496_test_tones_f4.flac with ODAC and Xonar Phoebus soundcards using three different headphones (KRK KNS-8400, Sennheiser HD 650 and AKG K601) and I heard a quiet high frequency tone in the IM parts with each setup with moderately loud volume. All these devices should perform quite nicely yet they all fail the IM part. I'm curious to know what kind of hardware does not.
Edit: ran RMAA tests on the soundcards in question. Xonar Phoebus headphone output @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/Xonar%20Phoebus%2096%20kHz.html) and ODAC into Xonar @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/ODAC%20-%20Phoebus%2096%20kHz.html)
I'm curious to know what kind of hardware does not.
Edit: ran RMAA tests on the soundcards in question. Xonar Phoebus headphone output @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/Xonar%20Phoebus%2096%20kHz.html) and ODAC into Xonar @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/ODAC%20-%20Phoebus%2096%20kHz.html)
NWAVguy measured the Behringer UCA202 (http://nwavguy.blogspot.com/2011/02/behringer-uca202-review.html) as having an impressive "IMD of 0.0009% is hard for anyone to argue with" although I don't know if his testing methodology correlates directly with RMAA specs.
I don't hear any faint tone at all through my UCA202 during his test section. [Except when I load the files into Audacity and listen to them there. Then I hear IM.]
That UCA202 doesn't seem too impressive based on a random RMAA test I found: link (http://rmaa.hege.li/UCA202.htm). For example ODAC with same sample rate measured with a worse sound card looks like this (http://www.saunalahti.fi/~cse/RMAA/ODAC.html).
Audacity has option to use WASAPI interface. That would be one explanation for the difference you experience. WASAPI playback would force 96 kHz output and the high frequencies and their problems are heard. If the other program uses DirectSound with Windows mixer set to 44.1 kHz or 48 kHz the excellent resampler in Windows would eliminate those high frequencies and issues would not be heard.
I don't mind comparing RMAA results when they are carried out with the same peripheral gear, by the same person/conditions, with the same levels, but comparing between Joe's RMAA results and Bill's RMAA results gets more sketchy. NWAVguy talks more about why here. (http://nwavguy.blogspot.com/2011/02/rightmark-audio-analyzer-rmaa.html)
Arny, last I heard, also has a UCA202, and hopefully can confirm if it appears as IM free as mine does.
I know there is also a UFO202 version with a phono preamp. I don't care about that part but the nice big ground post looks intriguing. I often have to unplug stuff (unused HDMI cord) to my laptop to eliminate minor ground loops. I wish I had a ground post on my UCA202 to play around with, perhaps it might help in some scenarios.
I should have checked the specs of that Behringer UCA202 immediately. It's only 48 kHz (http://www.behringer.com/EN/Products/UCA202.aspx) capable device so it doesn't really quality for this test. You are covered by Windows's excellent resampler and thus free from aliasing and distortion. Audacity uses its own resampler for playback and it defaults to low quality mode.
Testing 96 kHz playback requires device capable of that and unless ASIO or WASAPI is used Windows has to be configured properly too: (http://i.imgur.com/8kB9YfW.png).
OH NOES!! Worst news I've heard all year, but thanks for pointing that out Case.
Sorry Arny if I lead you astray on all your keys tests. Now, using my notebook's internal card, I can still show an ability to distinguish [post t=12.5s] but I definitely hear faint what I presume is IM [although it doesn't sound like your 4 kHz target tone sine wave, it is more raspy and buzzy, almost like a triangle or square wave, lower pitch, and perhaps dual tone] So I no longer qualify to even take the tests. *&^%$!
Here's me keying off this faint IM distortion, using my 96k setting of my Asus notebook's 3.5mm analog output, but of course keep in mind that may be EXACTLY what the organic twins are doing to too, even though they won't admit it:
foo_abx 2.0 report
foobar2000 v1.3.3
2015-04-02 23:49:44
File A: keys_jangling_full_band_2496_test_tones_f4.flac
SHA1: 5c0d71159fd3702d0515372876e699f1ca8de1d0
File B: keys_jangling_full_band_2496_1644E2Q150_2496_test_tones_f4.flac
SHA1: 6bbc99e2f0ca8f3083096b51fa0221792c6b5b85
Output:
DS : Primary Sound Driver
Crossfading: NO
23:49:44 : Test started.
23:51:20 : 01/01
23:51:24 : 02/02
23:51:28 : 03/03
23:51:31 : 04/04
23:51:35 : 05/05
23:51:38 : 06/06
23:51:42 : 07/07
23:51:45 : 08/08
23:51:45 : Test finished.
----------
Total: 8/8
Probability that you were guessing: 0.4%
-- signature --
c191c2e925611bd953a3489c4d4753e1213b0ebb
P.S. There are also audible differences in this IM distortion I hear when I change my notebook's advanced speaker settings output page from 96k/24 up to 192/24, should that mean anything.
Although I indeed hear IM issues now, they are faint and only from some PURE test tones, in isolation from other material, which are extremely high level compared to the actual keys sounds. It seems hard to believe that the keys noise itself, which doesn't sound any different between the two files, is actually being corrupted in an AUDIBLE manner which we need to worry about, that would not be immediately masked by the very broadband and constant sound they make on their own.* [I hope that makes sense. It is hard to verbalize.] I.e. you are shooting yourself in the foot by including the test tones at all. The keys should be a file all by themselves and if you want to tack on a side test for IM detection, so be it.
*It would be like worrying if a 4 kHz pure tone, some down 80 dB from the main signal, might be audible during the chaotic, constant sound of shattering glass. No, it will be masked.
No IM but the unfiltered version has a tiny click at the very end, so 3 while i hear only 2 with the filtered.
You'd love the IM I get on my oldest system here using Arny's sample. At moderate levels, the loudness of the IM is unrelated to the setting of the volume control! At volume=0, it's not there because the audio output is muted, but then between volume=1 and 10 (which is a range between almost inaudible and a reasonable listening level for normal audio signals) the loudness of the IM is pretty much constant.
This volume control is on a crappy Sony mini system which sits between the audio 2496 sound card and the HD 580 headphones.
Would you like me to tell you another secret that I learned a few years back? If you put ultrasonic audio content through something that's also processing analogue SD video, the relationships between the audio, and the harmonics of the video line frequency (15.625kHz, 31.25kHz etc for "PAL"), change the distortion products. Hence, if you don't remember there's also ~15kHz + harmonics already in the system, there will appear to be magic ultrasonic frequencies at which distortion increases and virtually disappears.
With all the facts, these things are quite predictable. Without all the facts, the distortion due to ultrasonic signal components can seem quite random and unpredictable. That's why it's good to see an IMD test signal with several different sets of frequencies in it.
Cheers,
David.
What output device do you have that doesn't have IM, Wombat?
Today I tested a few more soundcards I have. Realtek ALC1150 built-into the Maximus VI Formula motherboard, Creative Sound Blaster X-Fi Surround 5.1 Pro USB device, Creative Sound Blaster X-Fi HD USB device and Creative Sound Blaster X-Fi Titanium HD PCIe device all failed the test. The Titanium HD first appeared to pass the test but I found with RMAA that it was internally resampling to 48 kHz. Fixing that required installing Creative Console Launcher and with clock set to 96 kHz it failed the test too.
That's 6 tested DACs and none of them passed the test.
Another things to note is that the original sample has intersample peaks exceeding 0 dBFS causing clipping distortion with maximum volume on some of these devices. But lowering volume didn't remove the IM, it just made it quieter and required more attention to hear. Open headphones in an environment with a lot of background noise easily masked it but closed isolating headphones work wonders for hearing such detail.
I use the X-Fi Xtreme Music PCI in my gaming rig. W7 64 there, using audio modus in the Creative console set to 96kHz and also in the Win control panel. Open HD 590. No IM that i can spot. I never listen to loud and this jangling drives me nuts! Nothing i want to abx for long.
Would you be so kind to run RMAA test on the device? The X-Fi Titanium HD gets these (http://www.saunalahti.fi/~cse/RMAA/Creative%20SB%20X-Fi%20Titanium%20HD%20Headphone%20output%2096%20kHz.html) results at 96 kHz from the headphone output port.
Excuse the nonsense in device field. Seems like the latest RMAA version uses Russian too eagerly.
I should have checked the specs of that Behringer UCA202 immediately. It's only 48 kHz (http://www.behringer.com/EN/Products/UCA202.aspx) capable device so it doesn't really quality for this test. You are covered by Windows's excellent resampler and thus free from aliasing and distortion.
I certainly hope everyone is doing an RMAA test of their device's output
under the conditions they will be using it (e.g. same software settings and windows output APIs) before they do high frequency or ultrasound testing. If not, you really have no idea exactly what is being output.
Would you be so kind to run RMAA test on the device? The X-Fi Titanium HD gets these (http://www.saunalahti.fi/~cse/RMAA/Creative%20SB%20X-Fi%20Titanium%20HD%20Headphone%20output%2096%20kHz.html) results at 96 kHz from the headphone output port.
Excuse the nonsense in device field. Seems like the latest RMAA version uses Russian too eagerly.
I did some when reording my nvidia VGA cards a while back and it was ok but 48khz only. No interest atm for more.
No IM but the unfiltered version has a tiny click at the very end, so 3 while i hear only 2 with the filtered.
Good call. I hear that extra click at the end too: a dead obvious tell. [I think there is actually a click at the end of both but on one of them it is prominent, like the preceeding clicks, and on the other one it very faint.]
I'm now using HDMI out [I'm not a jitterphobe] at 96k to my prepro which verifies that the incoming signal is indeed 96, but I'm still plagued with faint IM.
Since I have been doing quite a bit of IM testing with Arnold's file I decided to isolate and fix the part I have been using. The attached file has nothing but the ultrasonic signals with smooth fades making it inaudible if there's no distortion.
[attachment=8253:im_test_tones.flac]
While at it I realized I had one more sound card capable of 96 kHz playback - Audioengine D1. Not surprising it failed the test too. Easiest way to prevent these distortions seems to be to let Windows resample everything to 44.1 or 48 kHz.
Arnold, I tested the keys_jangling_full_band_2496_test_tones_f4.flac with ODAC and Xonar Phoebus soundcards using three different headphones (KRK KNS-8400, Sennheiser HD 650 and AKG K601) and I heard a quiet high frequency tone in the IM parts with each setup with moderately loud volume. All these devices should perform quite nicely yet they all fail the IM part. I'm curious to know what kind of hardware does not.
Edit: ran RMAA tests on the soundcards in question. Xonar Phoebus headphone output @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/Xonar%20Phoebus%2096%20kHz.html) and ODAC into Xonar @ 96 kHz (http://www.saunalahti.fi/~cse/RMAA/ODAC%20-%20Phoebus%2096%20kHz.html)
I have noticed similar results with the Realtek ALC887 8-Channel High Definition Audio CODEC on the M5A97 II system board on this system. I have other audio interfaces and headphone preamps which I will attempt to test shortly.
The test tones at > 30 KHz are a more severe test than any pre-packaged test scheme such as the Audio Rightmark, that I know of. Not to fault good software because I think the Rightmark tests are just fine.
This whole test is basically very severe to the point of being irrational, and the levels and frequencies involved in both the keys jangling and test tones are worst case tests that are so severe as to be irrational as well.
Your listening test includes, but the RMAA testing does not include any IM that is created by the headphones themselves. I know of no known headphone IM testing at these frequencies, either.
This is of course a truism, but without actual acoustical measurements, who knows? The relevant measurement might be performed with a mic or SPL meter with normal 20-20 KHz response.
The headphones I am using are ATH-M50s and I have other headphones and earphones to test with.
The attached file has nothing but the ultrasonic signals with smooth fades making it inaudible if there's no distortion.
No more clicks to count, acting as a tell. Good job.
Your listening test includes, but the RMAA testing does not include any IM that is created by the headphones themselves. I know of no known headphone IM testing at these frequencies, either. This is of course a truism, but without actual acoustical measurements, who knows?
Sure there's IM in the electrical domain, in transducers, and in the acoustical domain, but what is freaky is there is something akin to IM in the human mind as well. Here's a test (http://upload.wikimedia.org/wikipedia/commons/a/a8/Binaural_beat_lossless_new.wav) which must be listened to on headphones. The left ear gets a pure, undistorted frequency and your right ear gets a slightly different frequency, however these two tones are in different channels so they can't combine to form IM products either in the electrical domain, your headphone driver, nor the acoustical domain, yet we hear a third product
in our minds: a binaural beat.
Your listening test includes, but the RMAA testing does not include any IM that is created by the headphones themselves. I know of no known headphone IM testing at these frequencies, either. This is of course a truism, but without actual acoustical measurements, who knows?
Sure there's IM in the electrical domain, in transducers, and in the acoustical domain, but what is freaky is there is something akin to IM in the human mind as well. Here's a test (http://upload.wikimedia.org/wikipedia/commons/a/a8/Binaural_beat_lossless_new.wav) which must be listened to on headphones. The left ear gets a pure, undistorted frequency and your right ear gets a slightly different frequency, however these two tones are in different channels so they can't combine to form IM products either in the electrical domain, your headphone driver, nor the acoustical domain, yet we hear a third product in our minds: a binaural beat.
Thanks for the information. So now there are two possible sources of audible IM that the Rightmark program cannot detect. However the test tones in the Keys Jangling test files is identically the same in both channels, so it would appear that binaural beating while interesting, is irrelevant.
Thanks for the information. So now there are two possible sources of audible IM that the Rightmark program cannot detect. However the test tones in the Keys Jangling test files is identically the same in both channels, so it would appear that binaural beating while interesting, is irrelevant.
Well that and its presumed that you can hear the primary frequencies in a binaural test. If its ultrasound, then the IMD has to happen somewhere earlier in the chain.
What's interesting about the binaural beat demo I linked to is it proves that in at least this particular instance, this extra sound, the "IM product" for lack of a better term, can't possibly be due to one's electronics, the transducers, etc.: it MUST be in the mind. My point though is that I believe there is also "perceived IM product" from dual tones* when presented to the SAME ear, created by the mind, however it is much more difficult to demonstrate this conclusively because we can't be 100% certain it isn't simply inherent to the signal which arrives at the ear drum, either due to, for instance, the IM of one's electronics or the IM of one's headphone driver, like we CAN be certain of in the binaural beats demo].
*Even from dual tones which are themselves ultrasonic!? Yikes! Hmm, that would be pretty freaky, I admit, but not impossible since we don't have a clear understanding of what causes the binaural beat phenomenon.
---
Why did you build in this IM test into your keys jangling challenge anyways, Arny? Was there some history of ultrasonic sounds causing audible band IM products, thereby contaminating the results, in some previous ultrasonics test? Oh wait, I'm now thinking of that wacky Oohashi (spelling?) guy and how the test that came after his which used alternate transducers for the different frequency bands showed no ability of listeners to differentiate the sounds with the "hypersonics" from the normal versions. Was that what prompted you?
I have now confirmed that the IM distortion indeed does come from the sound card. I connected the sound card's output to both headphones and line-in at the same time with Y-splitter cable and recorded what the device plays. The output looks like this:
(http://i.imgur.com/Rp6C7LK.png)
I then used foobar2000 to play the im_test_tones file looped and recorded the output from my headphones with iPhone and Dayton iMM-6 mic. It shows the same signals:
(http://i.imgur.com/D0tDUmc.png)
So as I read it the loudest component of the IM, with Case's soundcard, is 75 to 80 dB down from the test signal, which is itself markedly louder than the average level of the keys jangling. Do we really need to worry if the keys have some faint IM products of their own some 75 to 80 down from their level, especially considering their cacophonous and fairly wide band nature? I'm confident it will be perceptually masked.
If Arny were to release his keys challenge file without ANY test tones tacked on the end, I bet nobody, even me , will pass the test.
Since I have been doing quite a bit of IM testing with Arnold's file I decided to isolate and fix the part I have been using. The attached file has nothing but the ultrasonic signals with smooth fades making it inaudible if there's no distortion.
[attachment=8253:im_test_tones.flac]
While at it I realized I had one more sound card capable of 96 kHz playback - Audioengine D1. Not surprising it failed the test too. Easiest way to prevent these distortions seems to be to let Windows resample everything to 44.1 or 48 kHz.
Seems like I need to honor this worthy effort by including it into a distributed file collection.
So here you are:
[attachment=8254:keys_jan...tones_f5.flac]
[attachment=8256:keys_jan...tones_f5.flac]
About the test tones. Concerns about high frequency IM have been common to previous published attempts to isolate any possible audible differences due to increases in recording and playback system bandwidth. Loudspeakers with separate amplifiers and loudspeaker drivers for the range > 20 KHz are not uncommon.
It seems like people are more likely to report on audible differences in the test tones segment. This seems reasaonble and useful since it gives insights to the prevalence of audible problems with high frequency IM in real world playback system. But, such results need to be reported clearly and separately from positive results obtained with the actual keys jangling segment.
Based on previous experiments I suspect that spurious responses in the 20-44 KHz can be as small as -60 dB and positive results may still be reported, but not much smaller. This amount of nonlinear distortion (equivalent to 0.1% THD) does not seem to me to be an excessively demanding standard for modern audio gear.
Using my notebook's internal sound card at 96k, at somewhat elevated levels (but NOT clipping the target tone), keying on the IM sound of the test tones, source of IM unknown [which is exactly what the organic twins are probably doing but they won't admit it]:
foo_abx 2.0 report
foobar2000 v1.3.3
2015-04-06 20:18:59
File A: keys_jangling_full_band_2496_1644E2Q150_2496_test_tones_f5.flac
SHA1: 0c05bc4a9e99eb72528a4b12c6a2769638f234a0
File B: keys_jangling_full_band_2496_test_tones_f5.flac
SHA1: 1fde9c4e1c933827af0efcdba74cd10d0e2a88f1
Output:
DS : Primary Sound Driver
Crossfading: NO
20:18:59 : Test started.
20:19:31 : 01/01
20:19:36 : 02/02
20:19:43 : 03/03
20:19:47 : 04/04
20:19:51 : 05/05
20:19:57 : 06/06
20:20:01 : 07/07
20:20:06 : 08/08
20:20:06 : Test finished.
----------
Total: 8/8
Probability that you were guessing: 0.4%
-- signature --
0287eb7aa313132ae790c42cd2a8b82d4c9b685d
Using HDMI at 96k to my Marantz prepro in Pure Direct mode,
NOT at elevated mode at all, keying on the same IM distortion after the target tone, the difference is so obvious I simply "ran the Xs" and hardly ever even needed to hit A or B to reestablish my reference:
foo_abx 2.0 report
foobar2000 v1.3.3
2015-04-06 20:33:37
File A: keys_jangling_full_band_2496_test_tones_f5.flac
SHA1: 1fde9c4e1c933827af0efcdba74cd10d0e2a88f1
File B: keys_jangling_full_band_2496_1644E2Q150_2496_test_tones_f5.flac
SHA1: 0c05bc4a9e99eb72528a4b12c6a2769638f234a0
Output:
DS : Primary Sound Driver
Crossfading: NO
20:33:37 : Test started.
20:34:02 : 01/01
20:34:07 : 02/02
20:34:12 : 03/03
20:34:19 : 04/04
20:34:25 : 05/05
20:34:29 : 06/06
20:34:32 : 07/07
20:34:40 : 08/08
20:34:40 : Test finished.
----------
Total: 8/8
Probability that you were guessing: 0.4%
-- signature --
26523f2037de763acf8946ca2a296939942e0c5c
At least the audible clicks are now gone.
It seems like people are more likely to report on audible differences in the test tones segment. This seems reasaonble and useful since it gives insights to the prevalence of audible problems with high frequency IM in real world playback system. But, such results need to be reported clearly and separately from positive results obtained with the actual keys jangling segment.
Clearly? Separately? Hear, hear. Yet you insist on gluing the two segments together into one big file instead of keeping the keys test distinct from the IM test. Why?
Based on previous experiments I suspect that spurious responses in the 20-44 KHz can be as small as -60 dB and positive results may still be reported, but not much smaller. This amount of nonlinear distortion (equivalent to 0.1% THD
Incorrect. It may be around -60 dB down from full scale, which is close to where you put the test tones (just a tad below FS), sure, however the keys jangling part which is what we actually hear and we are focusing on (and use to set our playback levels)
you recorded some 30 to 40 dB down from full scale, so the net result is that this IM noise we hear from your tones is only 20 to 30 dB down from what we are actually listening to, the keys. And it appears in isolation after the keys so the keys can't mask it, making it dead obvious.
It seems like people are more likely to report on audible differences in the test tones segment. This seems reasaonble and useful since it gives insights to the prevalence of audible problems with high frequency IM in real world playback system. But, such results need to be reported clearly and separately from positive results obtained with the actual keys jangling segment.
Clearly? Separately? Hear, hear. Yet you insist on gluing the two segments together into one big file instead of keeping the keys test distinct from the IM test. Why?
I think the test tones would be ignored by most if not in the same file.
Based on previous experiments I suspect that spurious responses in the 20-44 KHz can be as small as -60 dB and positive results may still be reported, but not much smaller. This amount of nonlinear distortion (equivalent to 0.1% THD
Incorrect. It may be around -60 dB down from full scale, which is close to where you put the test tones (just a tad below FS), sure, however the keys jangling part which is what we actually hear and we are focusing on (and use to set our playback levels) you recorded some 30 to 40 dB down from full scale, so the net result is that this IM noise we hear from your tones is only 20 to 30 dB down from what we are actually listening to, the keys. And it appears in isolation after the keys so the keys can't mask it, making it dead obvious.
The peak level of the 24/96 keys jangling segment is very close to FS for me.
Left Right
Min Sample Value: -26307.07 -29561.82
Max Sample Value: 25997.06 24536.21
Peak Amplitude: -1.91 dB -.89 dB
Possibly Clipped: 0 0
DC Offset: 0 0
Minimum RMS Power: -72.17 dB -71.97 dB
Maximum RMS Power: -16.95 dB -16.93 dB
Average RMS Power: -30.03 dB -29.11 dB
Total RMS Power: -28.08 dB -27.17 dB
Actual Bit Depth: 24 Bits 24 Bits
Using RMS Window of 50 ms
Based on previous experiments I suspect that spurious responses in the 20-44 KHz can be as small as -60 dB and positive results may still be reported, but not much smaller. This amount of nonlinear distortion (equivalent to 0.1% THD
Incorrect. It may be around -60 dB down from full scale, which is close to where you put the test tones (just a tad below FS), sure, however the keys jangling part which is what we actually hear and we are focusing on (and use to set our playback levels) you recorded some 30 to 40 dB down from full scale, so the net result is that this IM noise we hear from your tones is only 20 to 30 dB down from what we are actually listening to, the keys. And it appears in isolation after the keys so the keys can't mask it, making it dead obvious.
The peak level of the 24/96 keys jangling segment is very close to FS for me.
The fleeting dynamic peaks, in this instance lasting for only tiny fractions of a second (and also hard to discern because they are extremely high frequency), may indeed be up there, but the
average level which we audibly focus on and use to set our volume knobs for comfortable playback is about 30 dB down from where the steady state IM test tones are recorded.
Average RMS Power: -30.03 dB -29.11 dB
I think the test tones would be ignored by most if not in the same file.
Perhaps you are an optimist and I am a pessimist. You designed the test from the perspective that fellow, scientifically minded people on (for instance) HA will take the test and honestly report if they hear IM, like I think everyone here (including myself now that I'm not resampling down via Windows) reports as having heard, despite trying a wide range of DACs. Me? I would have designed the test without inserting these tells which the organic twins, and their followers, can key on and then pretend "I heard a difference in the keys part. I swear.", which is what they are doing. You can't make a public challenge and tell people "Only concern yourself with the first 12 seconds and don't vote based on what you hear after that". They aren't going to follow that and instead are going to promote their Hi-Re$ mythology and pecuniary interests.
I also suspect that due to the loud, cacophonous, very dynamic nature of the keys' sound that any perception of faint, exceedingly brief moments of IM, which I think occur
only on the peaks, is unlikely. Studies which previously have tested using sustained test tones, similar to the ones you used, or other musical instruments, are very different from your keys' extremely brief peaks of ultrasonic content.
My theory that the ultrasonic content plummets if the fleeting peaks where to be removed could be tested by intentionally clipping the signal. If you have the means to do that I bet you'll see it effectively acts as a LPF, removing the ultrasonics.
Hearing faint IM products from your long, sustained test tones, in complete isolation from any other sounds which might mask our perception, is very different from transient IM produced by
, and during, the jangling keys.
Based on previous experiments I suspect that spurious responses in the 20-44 KHz can be as small as -60 dB and positive results may still be reported, but not much smaller. This amount of nonlinear distortion (equivalent to 0.1% THD
Incorrect. It may be around -60 dB down from full scale, which is close to where you put the test tones (just a tad below FS), sure, however the keys jangling part which is what we actually hear and we are focusing on (and use to set our playback levels) you recorded some 30 to 40 dB down from full scale, so the net result is that this IM noise we hear from your tones is only 20 to 30 dB down from what we are actually listening to, the keys. And it appears in isolation after the keys so the keys can't mask it, making it dead obvious.
The peak level of the 24/96 keys jangling segment is very close to FS for me.
The fleeting dynamic peaks, in this instance lasting for only tiny fractions of a second (and also hard to discern because they are extremely high frequency), may indeed be up there, but the average level which we audibly focus on and use to set our volume knobs for comfortable playback is about 30 dB down from where the steady state IM test tones are recorded.
Average RMS Power: -30.03 dB -29.11 dB
The keys jangling test signal is highly diagnostic for clipping, and this is an understatement!
Here is a sample with the gain increased 2 dB, clipped at FS and then dropped back by 2 dB, This is clipping of about 1.11 dB which is a tiny amount given that the crest factor of the file which is about 26 dB. The files peak at FS -0.89 dB. According to CEP 2.1 there are only 2 (two) clipped samples in the keys jangling portion of the file which has about 2.4 million samples in it. The file is thus clipped << 0.1% of the time.
[attachment=8257:keys_jan..._clip_f5.flac]
The above file is designed to be compared to: keys_jangling_full_band_2496_test_tones_f5.flac .
Here is my ABX log - no references to A or B during the test trials, just ran the X's:
foo_abx 2.0 beta 4 report
foobar2000 v1.3.5
2015-04-07 15:56:08
File A: keys jangling full band 2496 test tones f5.flac
SHA1: 1fde9c4e1c933827af0efcdba74cd10d0e2a88f1
File B: keys jangling full band 2496 test tones clip f5.flac
SHA1: 6a83beabc4cfd36af05c61ebaddcb20c0daeddca
Output:
DS : Primary Sound Driver
15:56:08 : Test started.
15:56:45 : 01/01
15:56:53 : 02/02
15:56:58 : 03/03
15:57:03 : 04/04
15:57:07 : 05/05
15:57:16 : 06/06
15:57:20 : 07/07
15:57:24 : 08/08
15:57:30 : 09/09
15:57:34 : 10/10
15:57:40 : 11/11
15:57:43 : 12/12
15:57:51 : 13/13
15:57:55 : 14/14
15:58:00 : 15/15
15:58:05 : 16/16
15:58:05 : Test finished.
----------
Total: 16/16
Probability that you were guessing: 0.0%
-- signature --
f1ac1deb470c882991da9006f52010ac6042888d
This was done while listening to just a 2 second segment of the keys jangling file t < 12 seconds. Hint: there are at least 2 different 2 second segments that are equally revealing.
If you want to use a picture of an impulse response as a means to estimate how audible ringing might be, then you should plot power (in dB) vs. time; e.g. see the graphs at the bottom of the page here: http://sox.sourceforge.net/SoX/Resampling (http://sox.sourceforge.net/SoX/Resampling)
The chart at the top of that page illustrates the conceptual problem. It portrays signals that are 90 dB down and are also > 20 KHz as if they are easily perceptible by means of making them as visible as signals at FS.
?? The point is that you seem to be trying to determine whether an LPF's ringing could possibly be audible by looking at its impulse response. This means being able to see when the ringing is perhaps 80dB down. So in the case of your roughly 35-pixel-high
linear-scale impulse response above, you need to be able to discern 0.0035 of a pixel:
(http://i60.tinypic.com/imjint.png)
Displaying the vertical scale for the impulse response as power in dB (
logarithmic) would remove this problem (don't know about CEP, but in Audacity this is just a simple click on a drop-down menu).
If you want to use a picture of an impulse response as a means to estimate how audible ringing might be, then you should plot power (in dB) vs. time; e.g. see the graphs at the bottom of the page here: http://sox.sourceforge.net/SoX/Resampling (http://sox.sourceforge.net/SoX/Resampling)
The chart at the top of that page illustrates the conceptual problem. It portrays signals that are 90 dB down and are also > 20 KHz as if they are easily perceptible by means of making them as visible as signals at FS.
?? The point is that you seem to be trying to determine whether an LPF's ringing could possibly be audible by looking at its impulse response. This means being able to see when the ringing is perhaps 80dB down. So in the case of your roughly 35-pixel-high linear-scale impulse response above, you need to be able to discern 0.0035 of a pixel:
(http://i60.tinypic.com/imjint.png)
Displaying the vertical scale for the impulse response as power in dB (logarithmic) would remove this problem (don't know about CEP, but in Audacity this is just a simple click on a drop-down menu).
I'm not trying to determine whether an LPF's ringing could possibly be audible by looking at its impulse response by means of analysis of objective evaluations. I'm trying to do proper listening tests to understand the situation better.
I'm simply pointing out that an existing tool seems to have a bias towards making a graphical mountain over what might be a perceptual mole hill.
Visual comparison of 7-14-2014 versions of the keys jangling files
[attachment=8258:keys_jan...sus_1644.png]
Please notice approximate 150 microsecond timing difference
For reference, the clearly audible 10 millisecond timing error between Mosaic A2 and Mosiac B2 from the AVS-AIX sample rate comparison files
[attachment=8260:Mosiac_a...ng_error.png]
Amd here is my evidence confirming that the difference is audible:
foo_abx 2.0 beta 4 report
foobar2000 v1.3.5
2015-04-08 19:24:01
File A: Mosaic_A2.wav
SHA1: 8ff31de028091c0daadb36c36bb51cbf88548e13
File B: Mosaic_B2.wav
SHA1: b899f3be763f92c642f7f6475f8cf3efad3023e7
Output:
DS : Primary Sound Driver
19:24:01 : Test started.
19:31:27 : 01/01
19:31:44 : 02/02
19:32:12 : 03/03
19:32:36 : 04/04
19:33:36 : 05/05
19:33:48 : 06/06
19:34:11 : 07/07
19:34:37 : 08/08
19:34:48 : 09/09
19:35:34 : 10/10
19:36:23 : 11/11
19:36:45 : 12/12
19:36:59 : 13/13
19:37:32 : 14/14
19:37:41 : 15/15
19:37:50 : 16/16
19:37:50 : Test finished.
----------
Total: 16/16
Probability that you were guessing: 0.0%
-- signature --
08f592c6e0dc1934c5435c20c95509a77000cd15
Methodology I used was to isolate a segment in the 3:24-3:34 range where the tell was most audible, and just switch back and forth between A and X and A and X until I heard the telltale echo, and if I failed there, or wished to confirm, did the same thing with B and X.
What output device do you have that doesn't have IM, Wombat?
There can be no such thing. Everything has nonlinear distortion, its just that for a lot of things like pure copper wire, its down quite a few dB.
The best I've seen is my AP 24192 that has all forms of distortion about 100 dB down over the 3 sets of test frequencies.
I don't know if i have much IM. I only know i don't hear it for these test samples at the loudnes i abx the keys.
The interesting thing is to what deggree IM is underrated as being a problem in all these comparisons.
We already wondered if this Meridian/AES test setup only was the most complicated setup ever to prove IM and not some filters.
I don't know if i have much IM. I only know i don't hear it for these test samples at the loudness I abx the keys.
That is the desired outcome - both files sound the same through their test tone region. They should sound the same and they should be silent.
The interesting thing is to what degree IM is underrated as being a problem in all these comparisons.
The purpose of this test adjunct is to let people find that out for themselves in exactly the same context as the rest of the test.
We already wondered if this Meridian/AES test setup only was the most complicated setup ever to prove IM and not some filters.
Good question!
IME the most common source of audible IM is clipping.
Meridian's CD bashing "typical" filter audibility paper made no claims to have tested for:
A) IM at the ear [could have been from their wacky, self amplified speakers, keep in mind, perhaps even in the transducers themselves let alone their electronics],
B) time synch accuracy [although refusing the listeners the ability to switch on the fly at their own will means this is less important, but not completely unimportant], and
C) precise level matching in the audible band
Even if they did use some level matching procedure, I bet they would have used something with, for instance, a simple K-weighted filter, like I believe Replaygain uses, which might very well be sensitive to ultasonics and therefore mis-matches the level in the audible, <20kHz range in order to be "correct" for the wideband average level.
Even if they did use some level matching procedure, I bet they would have used something with, for instance, a simple K-weighted filter, like I believe Replaygain uses, which might very well be sensitive to ultasonics and therefore mis-matches the level in the audible, <20kHz range in order to be "correct" for the wideband average level.
Whoa, this seems like a big problem to me.
Just threw together some samples in audacity of a 1kHz tone with amplitude 0.01, and in a second file I added a loud (0.8 amplitude) 24kHz tone:
[attachment=8261:0.01_amp...khz_tone.flac]
[attachment=8262:0.01_amp...khz_tone.flac] (warning: contains near full-scale ultrasonic tone!!!)
Replaygain (in foobar2k) reports wildly different track gains (+22 dB and -19 dB).
loudness-scanner (https://github.com/jiixyj/loudness-scanner), a command-line tool implementing EBU R128, reports -40.0 LUFS and 1.4 LUFS.
yet, they sound about the same volume, of course, because we can't hear the 24kHz tone.
I can see why you'd want a loudness measurement to alert you to loud ultrasonics in some contexts, but it seems to me that for level matching tracks in an ABX test, we really need a loudness measurement that takes into account the decreasing sensitivity of our hearing above ~15kHz, and totally ignores ultrasonics. i.e. respecting the right edge of these equal loudness curves: http://en.wikipedia.org/wiki/Loudness#/med...ile:Lindos1.svg (http://en.wikipedia.org/wiki/Loudness#/media/File:Lindos1.svg)
That wasn't such a big problem with the original ReplayGain
Though as I keep telling people, for near-transparent double-blind testing of objectively different audio signals, you really shouldn't use ReplayGain (any version) to match levels. It's not designed for that. Find some other way. e.g. knowing what the difference is in the audible band as part of the test design, and correcting for it.
Cheers,
David.
Luckily the latest version of foobar ABX shows in the summarized test results when ami r a user engages Replaygain [with its sensitivity to ultrasonics] in order to cheat, due to this very vulnerability I mentioned. Unfortunately this critical info was not part of the summarized report back when Arny first released his jangling keys test files in the AVS AIX records thread nor when the AIX records files themselves were released [making the posted test results of others rather questionable for all of these various ultrasonic tests, back then].
That wasn't such a big problem with the original ReplayGain
Though as I keep telling people, for near-transparent double-blind testing of objectively different audio signals, you really shouldn't use ReplayGain (any version) to match levels. It's not designed for that. Find some other way. e.g. knowing what the difference is in the audible band as part of the test design, and correcting for it.
When I'm preparing test files for ABXing, right up front I tack a few test tones in the normal audible range (20 Hz, 1 KHz, 16 KHz) onto the front or back of the file(s), process it, and then check to see if the test tones are still at the desired levels, signal purity, etc.
Nothing new about this - good recording, mixing and mastering engineers have been doing this for decades - nearly half a century. Doing this was particularly important when people were using Dolby Noise Reduction.
The following is a 22.0 KHz tone burst downsampled to 4416 CEP 2.1 Q=999 and upsampled back to 2496
[attachment=8263:2496_22_...16_Q_999.png]
The following is a 22.0 KHz tone burst downsampled to 4416 CEP 2.1 Q=150 and upsampled back to 2496
[attachment=8264:2496_22_...16_Q_150.png]
Some files for interchannel delay listening tests
[attachment=8278:impulses_2klp_4416.zip]
There is a larger collection of files, these are the highlights - reference. 1 sample, 9 samples, 27 samples
(1) A 0.45 second 24/192 file with a unit impulse in the middle was created
(2) It was appended 3 times to create a file with 4 impulses
(3) A number of copies were created with various delays added to the Left channel. 1,2,3,4,5,7,9,11,15,19,27 samples @ 192 kHz. Units of about 5 uSec
(4) All files were low pass filtered @ 2 KHz with a 10th order minimum phase low pass filter
(5) All files were down sampled to 4416
(6) all files were normalized to 90% (-1 dB FS)
Thanks. I'm going out now but will examine them more closely later.
edit: Preliminary ABX now posted in other thread and deleted from this post.
Some files for interchannel delay listening tests
Attached File
[attachment=8280:impulses...klp_4416.zip]
This is the upudated/corrected version. The 27 step delay file is designed to be very easy and give you an idea of what to listen to.
Test for 2 and 4 step delays were added to put in some more tests between doable and impossible.
1 sample delay = 5.2 uSec
2 sample delay = 10.4 uSec
4 sample delay = 20.8 uSec
9 sample delay = 46.85 uSec - This is roughly the result that David L. Clark reported.
27 sample delay = 140.6 uSec - this is a confidence builder - you should be able to complete it easily and accurately. It is also a test of the suitability of your test environment. Do not proceed to the shorter delays until you can do well with this file.
By the time you work your way down to the 1 step delay file, expect to sweat a little! ;-)
There is a larger collection of files, these are the highlights - reference. 1 sample delay, 2 samples delay, 4 samples delay, 9 samples delay, 27 samples delay.
Please start out with the 27 sample delay and work your way down.
This is how the files were created in CEP 2.1
(1) A 0.45 second 24/192 file with a unit impulse in the middle was created
(2) It was appended 3 times to create a file with 4 impulses
(3) A number of copies were created with various delays added to the Left channel. 1,2,3,4,5,7,9,11,15,19,27 samples @ 192 kHz. Units of about 5 uSec
(4) All files were low pass filtered @ 2 KHz with a 10th order minimum phase low pass filter
(5) All files were down sampled to 4416
(6) all files were normalized to 90% (-1 dB FS)
Attached files generated by:
for d in 0 1; do sox -V -n delay-${d}.flac synth 1s sq pad 100s 379s sinc -a 82 -t 4k -M 4k repeat 399 channels 2 delay ${d}s norm -1 trim 0 4; done
Attached files generated by:
for d in 0 1; do sox -V -n delay-${d}.flac synth 1s sq pad 100s 379s sinc -a 82 -t 4k -M 4k repeat 399 channels 2 delay ${d}s norm -1 trim 0 4; done
My analysis is that the file with the delay adds a 1 sample (22 uSec) delay to the left channel where the pulse rate is 100 Hz.
Can this methodology produce delays that are other than 1 sample?
Can this methodology produce delays that are less than 1 sample?
What are the typical ABX results?
Added test file with 3 sample shift. This file includes reference file and 5 previous test files;
[attachment=8289:impulses...klp_4416.zip]
This is the updated/corrected version. The 27 step delay file is designed to be very easy and give you an idea of what to listen to.
Test for 2 and 4 step delays were added to put in some more tests between doable and impossible.
1 sample delay = 5.2 uSec
2 sample delay = 10.4 uSec
3 sample delay = 15.6 uSec
4 sample delay = 20.8 uSec
9 sample delay = 46.85 uSec - This is roughly the result that David L. Clark reported.
27 sample delay = 140.6 uSec - this is a confidence builder - you should be able to complete it easily and accurately. It is also a test of the suitability of your test environment. Do not proceed to the shorter delays until you can do well with this file.
By the time you work your way down to the 2 sample delay file, expect to sweat a little! ;-)
There is a larger collection of files, these are the highlights - reference. 1 sample delay, 2 samples delay, 4 samples delay, 9 samples delay, 27 samples delay.
Please start out with the 27 sample delay and work your way down.
This is how the files were created in CEP 2.1
(1) A 0.45 second 24/192 file with a unit impulse in the middle was created
(2) It was appended 3 times to create a file with 4 impulses
(3) A number of copies were created with various delays added to the Left channel. 1,2,3,4,5,7,9,11,15,19,27 samples @ 192 kHz. Units of about 5 uSec
(4) All files were low pass filtered @ 2 KHz with a 10th order minimum phase low pass filter
(5) All files were down sampled to 4416
(6) all files were normalized to 90% (-1 dB FS)
[attachment=8298:1_KHz_ma...armonics.png]
Attached files generated by: for d in 0 1; do sox -V -n delay-${d}.flac synth 1s sq pad 100s 379s sinc -a 82 -t 4k -M 4k repeat 399 channels 2 delay ${d}s norm -1 trim 0 4; done
The sound is irritating, and seems right at my threshold based on Arny's test, so I didn't ABX these until just now. First attempt:
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-03 11:03:33
File A: delay-0.flac
SHA1: bf47cf35f523323d48e21e0b78b4627663c6021e
File B: delay-1.flac
SHA1: 8f34e5bc23ad74724b86162f768a1391939b7603
Output:
DS : Primary Sound Driver
Crossfading: NO
11:03:33 : Test started.
11:04:39 : 01/01
11:04:55 : 02/02
11:05:08 : 02/03
11:05:32 : 03/04
11:06:21 : 04/05
11:06:36 : 04/06
11:06:49 : 05/07
11:07:06 : 06/08
11:07:06 : Test finished.
----------
Total: 6/8
Probability that you were guessing: 14.5%
-- signature --
7a608ad88099aba9ecc6db3265ab8fda761576bb
Is this where I should be posting my results or some other thread?
I'll try again, I'm pretty sure I can ace this, but I have to wait for my refrigerator compressor to turn off so I can have dead silence, to concentrate.
edit to add: Second test had similar results. Third attempt here below. SCORE!
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-03 11:24:55
File A: delay-0.flac
SHA1: bf47cf35f523323d48e21e0b78b4627663c6021e
File B: delay-1.flac
SHA1: 8f34e5bc23ad74724b86162f768a1391939b7603
Output:
DS : Primary Sound Driver
Crossfading: NO
11:24:55 : Test started.
11:26:30 : 01/01
11:26:51 : 02/02
11:29:49 : 03/03
11:30:11 : 04/04
11:30:56 : 05/05
11:31:16 : 06/06
11:31:40 : 07/07
11:31:57 : 08/08
11:32:15 : 09/09
11:32:36 : 10/10
11:32:36 : Test finished.
----------
Total: 10/10
Probability that you were guessing: 0.1%
-- signature --
20caca8e62ab90c17b000f86cf8d8c607c9dedfe
P.S. I never in a million years would have thought that I could discern just a one sample shift. It seems too small. I'm suspicious there could be other elements such as artifacts I'm keying on. This is right on the cusp of my detection capabilities so I can't even say with 100% certainty what exactly it is I'm hearing that changes. Is it level? is it balance? is it tonality? Is it directional cues? Hmm, don't know. I'd say "directional cues" but maybe that's just from my observer bias kicking in?
When you guys make these files I'm thinking you shouldn't take a master file "A" and then present a secondary, delayed file version of it for "B", but unstead should take a master file, delay it by say 1 sample, call that "A", and then take the same master file and delay it 2 samples and call that one "B". This way both A and B have undergone processing, which may be introducing artifacts, but at least it will be the same for both A and B.
P.S. I never in a million years would have thought that I could discern just a one sample shift. It seems too small. I'm suspicious there could be other elements such as artifacts I'm keying on. This is right on the cusp of my detection capabilities so I can't even say with 100% certainty what exactly it is I'm hearing that changes. Is it level? is it balance? is it tonality? Is it directional cues? Hmm, don't know. I'd say "directional cues" but maybe that's just from my observer bias kicking in?
That's one of the beauties of doing DBTs with enough trials. When you hear something, even if it is too small to characterize, the statistics make it clear that you are hearing a diffference, and the controls ensure that you are hearing something.
I don't think that people who can't trust DBTs ever get into them enough to have this experience, and thus they miss out on a really convincing experience. I don't recall when we started experiencing this, but it was clearly happening 10 years after the first ABX tests.
When you guys make these files I'm thinking you shouldn't take a master file "A" and then present a secondary, delayed file version of it for "B", but unstead should take a master file, delay it by say 1 sample, call that "A", and then take the same master file and delay it 2 samples and call that one "B". This way both A and B have undergone processing, which may be introducing artifacts, but at least it will be the same for both A and B.
You already have some files with differential delays of that kind to listen to.
However, these particular files made by me are made by a particularly reliable procedure involving adding samples of silence to the front of one channel and not the other. The whole file is not processed by a delaying process, and the undelayed channel isn't processed by any delaying mechanism at all.
Arny, please post a 5 sample delay version of your recent tests. I will then attempt to ABX it not against the reference file but instead against the 1 sample delay version and then will know if what I am hearing is truly just a 4 sample delay difference or if your processing is introducing artifacts in the first test I took, the one where I heard a difference at the 4 sample, but not 3 sample, level of delay. Thanks.
[I also can attempt to compare 5 sample delay against 2 sample delay to see if for some unforeseen reason I can detect only a 3 sample difference that I couldn't before.]
Arny, please post a 5 sample delay version of your recent tests. I will then attempt to ABX it not against the reference file but instead against the 1 sample delay version and then will know if what I am hearing is truly just a 4 sample delay difference or if your processing is introducing artifacts in the first test I took, the one where I heard a difference at the 4 sample, but not 3 sample, level of delay. Thanks.
[I also can attempt to compare 5 sample delay against 2 sample delay to see if for some unforeseen reason I can detect only a 3 sample difference that I couldn't before.]
Since I'm on the road and don't have the original 24/192 impulses on my laptop, I'd like to put that off until I return home.
Yes, I could recreate them, but I might screw that up in some subtle way. So if you would be so kind as to wait a week... Pretty Please?;-)
While I'm cleaning that project up, you said something about background noise. I think I can make that go away. Necessary to fix?
Also, there might be a place in life for the whole file sequence to be recreated, only with a 2K High Pass filter in place of the 2K low pass filter. That would be closer to JJ's suggestion. Both need to be tested, and the compare & contrast analysis might be interesting.
Comments, anybody?
I looked at the Wayback Machine's files for PCABX.COM listening tests related to nonlinear distortion (the nonlinear piano) and found it to be missing some files and that other files were defective.
Here is hopefully a complete, correct set in FLAC format for your convenience:
[attachment=8304:piano_AM...s_FLACs_.zip]
The usual advice - start with the high order, high percentage distortion and work down.
Bold prediction: Everybody will hear one or more differences and in the end, some will be frustrated and others enlightened. ;-)
Yes, I could recreate them, but I might screw that up in some subtle way. So if you would be so kind as to wait a week... Pretty Please?;-)
Yes, that's fine. No rush.
While I'm cleaning that project up, you said something about background noise. I think I can make that go away. Necessary to fix?
Do you mean by adding dither noise to both A and B as a masker to the existing residual background noises? That would make the most sense to me. If not too much trouble, please do. If you mean something else, please run it by me first, and hopefully I'll understand what you are suggesting.
Yes, I could recreate them, but I might screw that up in some subtle way. So if you would be so kind as to wait a week... Pretty Please?;-)
Yes, that's fine. No rush.
While I'm cleaning that project up, you said something about background noise. I think I can make that go away. Necessary to fix?
Do you mean by adding dither noise to both A and B as a masker to the existing residual background noises? That would make the most sense to me. If not too much trouble, please do. If you mean something else, please run it by me first, and hopefully I'll understand what you are suggesting.
Umm those background noises are dither, unnecessarily raised by about 20 dB.
Here is hopefully a complete, correct set in FLAC format for your convenience: [attachment=8304:piano_AM...s_FLACs_.zip]
Your tenfold jumps in distortion level are a little drastic. Listening to only the odd harmonic ones, 3 and 5, I found 10% distortion to be dead obvious [ABX logs will be available upon request, should this be TOS 8 challenged] but 1% was impossible. Do you have intermediate 3% distortion files handy? Don't knock yourself out to make them if you don't.
Umm those background noises are dither, unnecessarily raised by about 20 dB.
Then what were you proposing to do to erase the difference in background noise I sometimes hear in your various tests? What I'm suggesting is to put the reference file through the exact same processing chain that you do for the secondary files, ie never let us listeners even get access to the original master file at all, only a processed version of it.
Here is hopefully a complete, correct set in FLAC format for your convenience: [attachment=8304:piano_AM...s_FLACs_.zip]
Your tenfold jumps in distortion level are a little drastic. Listening to only the odd harmonic ones, 3 and 5, I found 10% distortion to be dead obvious [ABX logs will be available upon request, should this be TOS 8 challenged] but 1% was impossible. Do you have intermediate 3% distortion files handy? Don't knock yourself out to make them if you don't.
That can be done: [attachment=8307:piano_AM_21_FLACs.zip]
[attachment=8306:piano_AM_21_FLACs.zip]
Umm those background noises are dither, unnecessarily raised by about 20 dB.
Then what were you proposing to do to erase the difference in background noise I sometimes hear in your various tests?
Just for the interchannel delay tests.
What I'm suggesting is to put the reference file through the exact same processing chain that you do for the secondary files, ie never let us listeners even get access to the original master file at all, only a processed version of it.
That is generally true of all of my samples.
I might do better with sustained flute notes, since they don't have such a congested rich harmonic structure like piano does, however I can't hear 3% either. Oh well. This seems slightly strange to me since 10% is so blatantly obvious.
I might do better with sustained flute notes, since they don't have such a congested rich harmonic structure like piano does, however I can't hear 3% either. Oh well. This seems slightly strange to me since 10% is so blatantly obvious.
Flutes tend to favor odd harmonics, and french horns seem to favor even harmonics. Am I going to have to produce two more sets of files?
Am I going to have to produce two more sets of files?
No need.
Out of curiosity, is the 5 second music excerpt from "In a restaurant by the sea" written by John Bucchino? Is this from some audio testing source analogous to the EBU SQAM files or did you just happen to pick it yourself?
Am I going to have to produce two more sets of files?
No need.
Out of curiosity, is the 5 second music excerpt from "In a restaurant by the sea" written by John Bucchino? Is this from some audio testing source analogous to the EBU SQAM files or did you just happen to pick it yourself?
The original piano sample came from an unidentified segment of music that was a part of a 24/96 demo by a company that made ADCs and DACs for professionals at the time.
Results of cubing test tones:
1 KHz test tone
[attachment=8308:Raw_cubed_1_KHz.png]
20 & 21 KHz twin tone
[attachment=8309:Raw_cube...0_21_kHz.png]
Screen shots during development of nonlinear distortion test files:
Contents of test file during development (not finished yet)
[attachment=8310:test_gen..._annoted.png]
Results of processing on 1 KHz sine wave
[attachment=8313:test_gen...KHz_test.png]
1 kHz = -0.441 dB 2 kHz -44.8 dB actual THD 0.12 %
Results of processing 20 and 21 KHz twin tone
[attachment=8311:test_gen...KHz_test.png]
Checking on possible aliasing due to nonlinear processing in digital domain. Also for checking eventual downsampling.
Results of processing frequency response test (swish test)
[attachment=8312:test_gen...ish_test.png]
0.3 dB down at 20 Hz deemed to be inconsequential as critical 50Hz-15 KHz range is easily within 0.1 dB. This LF roll off due to limited FFT size processing in CEP interacting with use of 192 KHz sampling for actual processing.
Origional file:
[attachment=8314:test_generation.png]
Even order file:
[attachment=8315:Raw_squared_.png]
Odd order file:
[attachment=8316:Raw_cubed_.png]
Admittedly these are dissimilar files, but doesn't the cubed file look much more like the original file than the squared file?
If you look at the 1 KHz tone as processed in the squared file, the fundamental has gone missing:
[attachment=8317:raw_squared_1_KHz.png]
If you look at the 20&21 kHz tones in the squared file, they have gone missing as well:
[attachment=8318:raw_squa...0_21_kHz.png]
Not so with the cubed file. All the original test tones are represented, albeit at somewhat changed but still highly significant levels.
For example, the 1 KHz segment of the cubed file:
[attachment=8319:Raw_cubed_1_KHz.png]
Transformer-Based Speaker Simulators
[attachment=8322:speaker_simulators.png]
Measuring headphone jack source impedance
[attachment=8326:impedanc...IMG_0903.png]
Conference says:
"Upload failed. The file was larger than the available space"
This is a lie. The file is 6.1 megs. The available space is supposed to be 11.8 megs. New math? ;-)
My apologies. I went back to the same problem after a few reboots of windows 8.1 and now the troublesome zip file was 12.8 megs. The conference was right!
So I shortened the files.
These are the 24/96 ISC (Inter-Sample Clipping) test files that allow one to hear how files that contain no clipped samples can still sound clipped.
[attachment=8330:keys_jan..._shorter.zip]
Usual caveats. This is basically the keys jangling file with lots of ultrasonic content, so be careful when turning the volume up. Delicate tweeters could be harmed, especially if your hearing is a little deficient at the high frequency end. Headphones are recommended.
There are 3-4 instances of ISC clipping involving a very small number of samples distributed at approximately 2-3 second intervals throughout the file.
The file contains 9.5 seconds of the classic 24/96 keys jangling sound with and without intersample clipping. They are level matched, etc. and need no further processing.
Following the keys jangling sound is a low amplitude 1 second 4 KHz test tone to show that the primary listening test is over and the secondary listening test of the linearity of your monitoring system has begun.
You can use the FB2K file controls to keep from running into this part of the test until you have completed the primary test.
As you go back and forth between the two files from this point through the end of the file you should hear either nothing or a very low level rushing sound. Any other sound or audible difference between the two files indicates that your monitoring system is not of sufficient quality to be unconditionally undistorted when operated at 24/96.
I did do a FB2K listening test with 16/16 results but FB2K crashed while I was saving the log file.
It is not intersample clipping one hears in the test file but soundwave discontinuity making a nasty pop sound. Same sound is present even when amplitude is lowered by several dBs more than the intersample peaks above full scale.
It is not intersample clipping one hears in the test file but soundwave discontinuity making a nasty pop sound. Same sound is present even when amplitude is lowered by several dBs more than the intersample peaks above full scale.
Please describe the means you used to "lower the amplitude".
Initially in foo_abx using the volume control in foobar2000. That works in digital domain and would remove clipping if it were present. Then opened the file in Adobe Audition CS6 and used Effects -> Amplitude and Compression -> Amplify... -> and set Gain to -15 dB for both channels.
Initially in foo_abx using the volume control in foobar2000. That works in digital domain and would remove clipping if it were present.
That would be true if the function in question was immune to the effects of digital clipping. How do we know that this is true?
Then opened the file in Adobe Audition CS6 and used Effects -> Amplitude and Compression -> Amplify... -> and set Gain to -15 dB for both channels.
Now that is convincing evidence!
I just did something similar and obtained similar results. I have therefore confirmed your results by means of a far simpler path. Just attentuate the file and play it by the simplest possible means.
Thank you for pointing out this fatal error in my first attempt at demonstrating this problem.
I was searching about for some means to validate the test and I thank you for your insights.
Back to the "Drawing Board".
This is a updated/corrected version. The 27 step delay file is designed to be very easy and give you an idea of what to listen to.
[attachment=8333:impulses...rm_4416_.zip]
Test for 5, 7, 9 and 15 step delays were added to put in some more tests between doable and impossible.
The noise floor has been dropped dramatically by simply changing up the order of the last 2 operations.
The 0 sample delay file is your reference file - compare all other files to it.
1 sample delay = 5.2 uSec
2 sample delay = 10.4 uSec
3 sample delay = 15.6 uSec
4 sample delay = 20.8 uSec
5 sample delay = 26.0 uSec
7 sample delay = 36.4 uSec
9 sample delay = 46.85 uSec - This is roughly the result that David L. Clark reported.
15 sample delay = 78.0 uSec
27 sample delay = 140.6 uSec - this is a confidence builder - you should be able to complete it easily and accurately. I do 16/16. It is also a test of the suitability of your test environment.
Please do not proceed to the shorter delays until you can do well with the 27 step file.
By the time you work your way down to the 2 sample delay file, expect to sweat a little! ;-)
Please start out with the 27 sample delay and work your way down.
This is how the files were created in CEP 2.1
(1) A 0.45 second 24/192 file with a unit impulse in the middle was created
(2) It was appended 3 times to create a file with 4 impulses
(3) A number of copies were created with various delays added to the Left channel. 1,2,3,4,5,7,9,11,15,19,27 samples @ 192 kHz. Units of about 5 uSec
(4) All files were low pass filtered @ 2 KHz with a 10th order minimum phase low pass filter
(5) all files were normalized to 90% (-1 dB FS)
(6) All files were down sampled to 4416
Changing the order of 5 & 6 dramatically dropped the noise floor.
Haven't tried 3 sample delay yet but 4 is doable with a nice quiet room. Unfortunaely there is traffic, trucks, and refrigerator noise here I have to contend with.
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-27 13:08:01
File A: Impulses shift 4 samples 2klp norm 4416 .flac
SHA1: 6133aaa124c97a3f768f3d9216af2eb07b7c0bf3
File B: Impulses shift 0 samples 2klp norm 4416 .flac
SHA1: 8fc00a4bb6a1bb0a66ec5c83cfaa36f9d8fddd13
Output:
DS : Primary Sound Driver
Crossfading: NO
13:08:01 : Test started.
13:08:56 : 00/01
13:09:11 : 01/02
13:09:33 : 02/03
13:09:46 : 03/04
13:10:11 : 04/05
13:10:32 : 05/06
13:11:10 : 06/07
13:11:20 : 07/08
13:11:47 : 08/09
13:12:07 : 09/10
13:12:07 : Test finished.
----------
Total: 9/10
Probability that you were guessing: 1.1%
-- signature --
b5a468b183d7a98c037ae06ef08563400d89231b
Arny, are you sure David Clark's tests were with headphones, not speakers?
3 takes more effort. My first stab at it, without training, was pretty much random but my second test was not as bad:
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-27 14:03:20
File A: Impulses shift 3 samples 2klp norm 4416 .flac
SHA1: 6fee6af7eedd131b9809333f86e5194539ca2f63
File B: Impulses shift 0 samples 2klp norm 4416 .flac
SHA1: 8fc00a4bb6a1bb0a66ec5c83cfaa36f9d8fddd13
Output:
DS : Primary Sound Driver
Crossfading: NO
14:03:20 : Test started.
14:04:49 : 01/01
14:05:49 : 02/02
14:06:13 : 02/03
14:06:28 : 03/04
14:06:46 : 04/05
14:07:49 : 05/06
14:08:17 : 06/07
14:08:34 : 07/08
14:09:22 : 07/09
14:10:25 : 08/10
14:10:25 : Test finished.
----------
Total: 8/10
Probability that you were guessing: 5.5%
-- signature --
9f78862d1c5ea5ba107f26fe2337a82ae1778651
2 is very hard.
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-27 14:19:43
File A: Impulses shift 2 samples 2klp norm 4416 .flac
SHA1: 0d55d7f19f67bcf33b9ca9416e9e7981e098ed03
File B: Impulses shift 0 samples 2klp norm 4416 .flac
SHA1: 8fc00a4bb6a1bb0a66ec5c83cfaa36f9d8fddd13
Output:
DS : Primary Sound Driver
Crossfading: NO
14:19:43 : Test started.
14:22:05 : 00/01
14:22:28 : 01/02
14:22:46 : 02/03
14:23:18 : 03/04
14:24:29 : 04/05
14:25:44 : 05/06
14:26:35 : 06/07
14:26:58 : 07/08
14:27:37 : 08/09
14:28:14 : 08/10
14:28:14 : Test finished.
----------
Total: 8/10
Probability that you were guessing: 5.5%
-- signature --
852b926dbe8de4708857e9b624e2f167e8d621fa
My perception is that I switch from lateral soundstage localization cues to instead using tonal balance differences. One sound is more of a "thump" and the other is more of a "thunk" sound.
My understanding is that when we reach the JND points it is not at all uncommon for the listener to lose sight of what exactly it is that they are keying on. Is it level? Tone? Direction? Other?
edit to add: Just tried 1 sample shift for yucks. No go.
Haven't tried 3 sample delay yet but 4 is doable with a nice quiet room. Unfortunaely there is traffic, trucks, and refrigerator noise here I have to contend with.
foo_abx 2.0 report
foobar2000 v1.3.3
2015-05-27 13:08:01
File A: Impulses shift 4 samples 2klp norm 4416 .flac
SHA1: 6133aaa124c97a3f768f3d9216af2eb07b7c0bf3
File B: Impulses shift 0 samples 2klp norm 4416 .flac
SHA1: 8fc00a4bb6a1bb0a66ec5c83cfaa36f9d8fddd13
Output:
DS : Primary Sound Driver
Crossfading: NO
13:08:01 : Test started.
13:08:56 : 00/01
13:09:11 : 01/02
13:09:33 : 02/03
13:09:46 : 03/04
13:10:11 : 04/05
13:10:32 : 05/06
13:11:10 : 06/07
13:11:20 : 07/08
13:11:47 : 08/09
13:12:07 : 09/10
13:12:07 : Test finished.
----------
Total: 9/10
Probability that you were guessing: 1.1%
-- signature --
b5a468b183d7a98c037ae06ef08563400d89231b
Arny, are you sure David Clark's tests were with headphones, not speakers?
Sorry if I conveyed the impression that Clark's ancient tests were done with headphones. Not true. It's been a while but my recollection is that the test rig was two midrange speakers on tracks that slid closer and farther from the listener, and the distance difference produced a delay that was part of the test. I think there may have been more to it than that, but it was definitely 2 speakers operated like that.
Looks like the cleaned up samples are giving you better results, no?
Sensitivity using speakers instead of headphones should be lower so that would explain why my old ears are outdoing his test subjects.
Each time I take your tests I'm also getting training, so my better results may be due to that, I'm not sure. I wish other people would post their results so we could compare.
Clark's test for ITD by moving speakers forwards and backwards is quite poor if you ask me, even if he compenstated for L vs. R level changes. Besides delay, using the method you described, he would have gotten differences in L vs R level, the acoustical power response measured at the ear would be different [nearfield response is dictated more heavily by direct, on-axis response and farfield sound is more heavily influenced by the summed reflected sound], plus there will be changes to the constructive/destructive comb filtering when any two speakers are separated by differing distances.
Sensitivity using speakers instead of headphones should be lower so that would explain why my old ears are outdoing his test subjects.
My standard for comparison is my recollection of the results that you obtained with the last (noisy) set of samples. There was an outstanding request for a test involving what, 6 samples as that was around the place where you seemed to stop hearing differences. That's my recollection.
Now you are mucking around 2 samples. I'm coming up for my plan to do test files for < 1 sample. ;-)
Each time I take your tests I'm also getting training, so my better results may be due to that, I'm not sure.
Listener training works? That's not news! ;-)
I wish other people would post their results so we could compare.
I'd love to have other people try because the results of one listener don't have much weight.
I'd also like to have the samples themselves checked out for false positives, other errors.
Hopefully there will be people who are interested. I think that very non-golden means have been used to produce fairly golden results.
I'd love to have other people try because the results of one listener don't have much weight.
Listening to impulse sounds is pure torture to me. But with volume turned to barely audible level I tested the 27, 15 and 9 sample delay files. Only thing I hear is that the sound source seems to move more towards left the more delay there is. I did a quick test with the single sample delay file too but couldn't notice any change there.
Is there a reason these tests, like the one bandpass posted he made with sox, are using impulses? It's the most horrible sound one can imagine.
I'd love to have other people try because the results of one listener don't have much weight.
Listening to impulse sounds is pure torture to me. But with volume turned to barely audible level I tested the 27, 15 and 9 sample delay files. Only thing I hear is that the sound source seems to move more towards left the more delay there is. I did a quick test with the single sample delay file too but couldn't notice any change there.
That actually seems pretty good. Thank you!
Is there a reason these tests, like the one bandpass posted he made with sox, are using impulses? It's the most horrible sound one can imagine.
The reason why we use simple synthetic sounds in tests like these is that we are trying to use signals that are in some sense like some parts of music, but lack the complexity of music that causes so much masking and vastly reduces the ears sensitivity to small differences.
In most cases small differences like these are not heard on an ongoing basis while listening to music, dialog or drama.
Instead there are Critical Passages where the sounds coincide in a special way, and bang, it becomes noticeable that something is not right.
Finding those Critical Passages can take a lot of time, and so some of us try to short-cut the process by coming up with artificial sounds that based on what we already know about hearing and listening, are in some sense like the natural sounds, and cause the same technical flaws to be audible.
In actual music, we've found that impulsive sounds like castanets can be good for finding flaws like these. Castanets generate impulses that are actually fairly pure, but with some bandpass filtering and ringing tossed in due to the way they are made. They are probably the musical instrument that spawned a lot of tests like these.
This is a updated/corrected version. The 27 step delay file is designed to be very easy and give you an idea of what to listen to.
[attachment=8333:impulses...rm_4416_.zip]
Test for 5, 7, 9 and 15 step delays were added to put in some more tests between doable and impossible.
The noise floor has been dropped dramatically by simply changing up the order of the last 2 operations.
The 0 sample delay file is your reference file - compare all other files to it.
Mention of these files on the HA forum yielded 2 downloads, one of which was my test of it.
Mention of these files on the Squeezebox forum about a week ago yielded 0 downloads.
Mention of these files on the Head Fi forum yielded 5 downloads and it was just yesterday.
I don't look at those other forums. If anyone beats my ABX score of 2 sample delay discrimination [with a p val. .055] , let me know.
[edit to add.] Considering I did much better than Clark's findings you mentioned, I looked into it. My 10 microsecond level isn't so special (http://web.mit.edu/hst.722/www/Topics/Quantitative/Shackleton2003.pdf).
"The smallest change in the ITD of pure tones detectable by humans
[just noticeable difference (jnd)] is 10–20 microsec (Mills, 1958;
Durlach and Colburn, 1978; Hafter et al., 1979)"
I don't look at those other forums. If anyone beats my ABX score of 2 sample delay discrimination [with a p val. .055] , let me know.
[edit to add.] Considering I did much better than Clark's findings you mentioned, I looked into it. My 10 microsecond level isn't so special (http://web.mit.edu/hst.722/www/Topics/Quantitative/Shackleton2003.pdf).
"The smallest change in the ITD of pure tones detectable by humans
[just noticeable difference (jnd)] is 10–20 microsec (Mills, 1958;
Durlach and Colburn, 1978; Hafter et al., 1979)"
I am perfectly satisfied if we can reaffirm the classic thresholds of audibility with ABX testing. How could we expect to do better?
I am perfectly satisfied if we can reaffirm the classic thresholds of audibility with ABX testing. How could we expect to do better?
Well in some instances my perception is clearly slight alterations in lateral positioning, just as we would expect, however in other instances I suspect my brain hears the two impulses and reinterprets them as two crests of a wave form, two cycles long in duration, and hears them as slightly different
frequencies. [Sort of a singular "thunk" vs "thump/thud" sound.] This second detection method is not what we are testing for, ITD, but I can't stop my brain from using it as a "cheating method" now can I?
Although I can see how using impulses in isolation would be useful to increase sensitivity and discrimination even down to the 10 microsecond level, or so, I fear there may be other things at play such as frequency detection that I'm subconsciously keying on. Listening instead to music or perhaps even better, continuous correlated pink noise, would be a possible solution. [I can't hear a "frequency" to the thump noise when there
is no thump!]
I never would have thought it, but I know I can hear ITD even down to the small single digits of milliseconds when listening to correlated pink noise. I'm not sure if I can get down to the microsecond levels but I would be more than glad to try if you post it. To the best of my knowledge using this would cure the issue I'm worrying about.
That's not cheating, and your brain does combine the "signals" from both ears at some level. Google binaural beats.
Cheers,
David.
Well I'm not sure if the word is "cheating" however the part of my brain which does traditional ITD analysis and creates a mental picture of lateral sound stage image localization is no longer being used when I switch into frequency analysis mode and the direction the single "thunk" sound seems to come from does not change, just its tonality.
The physical property that has changed is the ITD. If you can detect the change, you can detect the change. It doesn't matter how your brain is doing it. If it's detected, it's above threshold. If you can't detect it at all, it's below threshold.
IMO
Cheers,
David.
I see it as an unintentional artifact of his chosen source material. If he instead used correlated pink noise (mono) then there are no pairs of impulses for the listener to mentally combine together to form a telltale frequency.
Think of that terrible way Clark did it where he moved two speakers forwards and backwards to change the ITD. Terrible! Whenever you have two drivers in the same room their differing spacing will make changes in their comb filtering from differences in the constructive/destructive soundwave summing of the two sources. Who knows if people were keying off of that or what they were supposed to, ITD.
I agree that what you describe isn't just testing ITD.
However, clicks or tones or noise through headphones is. There may be different thresholds for different stimuli, which is interesting in itself, but it's still fair for a given stimuli.
Cheers,
David.
I see it as an unintentional artifact of his chosen source material. If he instead used correlated pink noise (mono) then there are no pairs of impulses for the listener to mentally combine together to form a telltale frequency.
Think of that terrible way Clark did it where he moved two speakers forwards and backwards to change the ITD. Terrible! Whenever you have two drivers in the same room their differing spacing will make changes in their comb filtering from differences in the constructive/destructive soundwave summing of the two sources. Who knows if people were keying off of that or what they were supposed to, ITD.
I talked to Clark about it Tuesday night over pulled pork, and there was only 1 speaker and the delays were introduced electronically.
How do you test interaural time delay with one speaker? I think you/he are misremembering something.
How do you test interaural time delay with one speaker? I think you/he are misremembering something.
You don't and that was a signficant difference between our tests.
My test was an inte rpulse time delay test that became a inter aural time delay test when played through headphones.
If one summed the channels of my test, you came close to Clark's tests.
Clark is not exactly a fan of headphones, and he made his living working with speakers.
NEW
24/96 versus 1644 files with more different kinds of musical samples:
[attachment=8364:dac_test...496_dist.flac]
[attachment=8365:dac_test...96_distf.flac]
There are 4 different selections, the first 3 are musical and the last is the usual keys jangling
You can use the ABX plug in in FOOBAR2K to restrict your listening to any part of any of them, so a minimum of 4 different tests are possible.
There is a ultrasonic IM test at the end after the short little audible test tone. Make sure that you can't hear the difference between the 44 and 96 KHz versions. You may have to turn your listening level down a few dB to obtain the desired totally silent results.
Other than the usual vulnerability of the test tone section (which people like the organic twins will surely exploit without admitting to it)
foo_abx 2.0.1 report
foobar2000 v1.3.3
2015-08-03 10:54:15
File A: dac_test_1644_2496_dist.flac
SHA1: d27e428fc69429ed7a22ed3e650a60fed2d6e246
File B: dac_test_2496_distf.flac
SHA1: 79d74a1262091a2ecf707a6e267c2c4a451f0459
Output:
DS : Primary Sound Driver
Crossfading: NO
10:54:15 : Test started.
10:56:50 : 01/01
10:56:55 : 02/02
10:56:59 : 03/03
10:57:08 : 04/04
10:57:12 : 05/05
10:57:16 : 06/06
10:57:19 : 07/07
10:57:23 : 08/08
10:57:27 : 09/09
10:57:30 : 10/10
10:57:30 : Test finished.
----------
Total: 10/10
Probability that you were guessing: 0.1%
-- signature --
470735494f3cca9d6a3c5b21df1fdbc419c0f02e
there's no audible difference.
Other than the usual vulnerability of the test tone section (which people like the organic twins will surely exploit without admitting to it)
foo_abx 2.0.1 report
foobar2000 v1.3.3
2015-08-03 10:54:15
File A: dac_test_1644_2496_dist.flac
SHA1: d27e428fc69429ed7a22ed3e650a60fed2d6e246
File B: dac_test_2496_distf.flac
SHA1: 79d74a1262091a2ecf707a6e267c2c4a451f0459
Output:
DS : Primary Sound Driver
Crossfading: NO
10:54:15 : Test started.
10:56:50 : 01/01
10:56:55 : 02/02
10:56:59 : 03/03
10:57:08 : 04/04
10:57:12 : 05/05
10:57:16 : 06/06
10:57:19 : 07/07
10:57:23 : 08/08
10:57:27 : 09/09
10:57:30 : 10/10
10:57:30 : Test finished.
----------
Total: 10/10
Probability that you were guessing: 0.1%
-- signature --
470735494f3cca9d6a3c5b21df1fdbc419c0f02e
there's no audible difference.
????
Looks to me like there was an audible difference and you nailed it.
Where was the audible difference?
The audible difference was during the test tone sequence. The faint background noise changes. I guess you could argue it is IM on either my end or your end but boy is it faint.
If you were to prepare two dead silent tracks using the exact same methodology, with no signal at all, this would tell us if the change in noise floor I can pick up on occurs also in the absence of test tones.
The audible difference was during the test tone sequence. The faint background noise changes. I guess you could argue it is IM on either my end or your end but boy is it faint.
If you were to prepare two dead silent tracks using the exact same methodology, with no signal at all, this would tell us if the change in noise floor I can pick up on occurs also in the absence of test tones.
I had that problem and addressed it by turning down the playback level a little.
It looks like your monitoring environment may be faulty.
Two dead silent tracks would seem to shed very little light. They would both be silent. What would cause them to be different?
I had that problem and addressed it by turning down the playback level a little. It looks like your monitoring environment may be faulty.
Since your monitoring environment seems to be faulty as well, how is it we know for sure that the problem isn't embedded in the files themselves?
Dead silent tracks would allow me to determine if what I hear is noise or signal distortion. If there
is no signal, yet I still hear a difference, then what I'm hearing can't be distortion, it is noise.
I had that problem and addressed it by turning down the playback level a little. It looks like your monitoring environment may be faulty.
Since your monitoring environment seems to be faulty as well, how is it we know for sure that the problem isn't embedded in the files themselves?
I was speaking of that particular monitoring environment, not the only monitoring environment that I have. I also have a 20 MHz dual trace oscilloscope to use to visualize clipping.
However, any real world monitoring environment is going to have finite dynamic range, so if the signal is loud enough and there are adequate gain reserves, the headphone amp can be possibly pushed in to clipping.
No matter what means are used to alleviate any problems with the monitoring system test, if they work, then they are fitting and proper.
Dead silent tracks would allow me to determine if what I hear is noise or signal distortion. If there is no signal, yet I still hear a difference, then what I'm hearing can't be distortion, it is noise.
If the track is dead silent, then there is no signal to stimulate distortion.
If the track is dead silent, then there is no signal to stimulate distortion.
Excellent. So you agree, if I hear the same audible difference with dead silence then that would eliminate any notion that what causes the audible distinction is general IM. Or for that matter clipping. All I would be comparing is the noise floors. This would interest me, but seeing as generating two such files is a complex, laborious process on your part I can see why you are scoffing at providing them.
If the track is dead silent, then there is no signal to stimulate distortion.
Excellent. So you agree, if I hear the same audible difference with dead silence then that would eliminate any notion that what causes the audible distinction is general IM. Or for that matter clipping. All I would be comparing is the noise floors. This would interest me, but seeing as generating two such files is a complex, laborious process on your part I can see why you are scoffing at providing them.
Its not too laborious for me to cut to the chase, I think. ;-)
I think this is what you want.
[attachment=8367:digital_..._vs_4416.flac]
The above file is a 24/96 flac file. The first 10 seconds were made by simply generating 24/96 digital black, there is a impulse, and there are 10 seconds of 24/96 digital black downsampled to 16/44 and upsampled back to 24/96.
They should sound the same. There is no signal so none of the usual forms of nonlinear distortion are involved.
What do you hear?
But there's no way for me to test it without the potential problem of observer bias. I'm kindly requesting two files, not one, so I can attempt to ABX them against each other, of digital black. Adding impulse noises, keys jangling, or music before or after the 10 second digital black part is fine by me if you insist, however should I successfully distinguish one digital black from the other you'll never know for sure if it was actually by my listening to the music/impulse/keys jangling part that allowed me to distinguish A from B or if it was just a difference in the noise floors. Add the extra content, besides just the digital black part I'll be focusing on, if you wish.
The above file is a 24/96 flac file. The first 10 seconds were made by simply generating 24/96 digital black, there is a impulse, and there are 10 seconds of 24/96 digital black downsampled to 16/44 and upsampled back to 24/96.
This, for example, will work as one of the two files I'm requesting. I need a second version of this please, which never had any downsampling and then upsampling section: it was kept as 24/96 the entire time. Then I can attempt to ABX the two files. Thanks.
But there's no way for me to test it without the potential problem of observer bias. I'm kindly requesting two files, not one, so I can attempt to ABX them against each other, of digital black. Adding impulse noises, keys jangling, or music before or after the 10 second digital black part is fine by me if you insist, however should I successfully distinguish one digital black from the other you'll never know for sure if it was actually by my listening to the music/impulse/keys jangling part that allowed me to distinguish A from B or if it was just a difference in the noise floors. Add the extra content, besides just the digital black part I'll be focusing on, if you wish.
The above file is a 24/96 flac file. The first 10 seconds were made by simply generating 24/96 digital black, there is a impulse, and there are 10 seconds of 24/96 digital black downsampled to 16/44 and upsampled back to 24/96.
This, for example, will work as one of the two files I'm requesting. I need a second version of this please, which never had any downsampling and then upsampling section: it was kept as 24/96 the entire time. Then I can attempt to ABX the two files. Thanks.
OK and since file length of files like these are short or vanishing, they are both now 60 seconds for relaxed ABXing:
[attachment=8368:digital_...ack_2496.flac] digital black - generated as 2496, not resampled
[attachment=8369:digital_...644_2496.flac] digital black - generated as 2496, resampled to 1644 and back to 2496.
The actual question is whether one can hear the highly shaped dither.
[attachment=8370:2496_dig..._ds_1644.png]
I just installed Windows 10 today and I'm trying to work out some bugs. Luckily FB2K ABX seems to still work.
I'm going to sleep now but here's the log. I'll add more commentary tomorrow.
foo_abx 2.0.1 report
foobar2000 v1.3.3
2015-08-06 02:18:08
File A: digital_black_1644_2496.flac
SHA1: 5f1bdf3767076523d2065835e67f43a6ce08d580
File B: digital_black_2496.flac
SHA1: e8f61474e2fc7628ad3d276d0fb0f3ad61d53d98
Output:
DS : Primary Sound Driver
Crossfading: NO
02:18:08 : Test started.
02:20:42 : 01/01
02:20:49 : 02/02
02:20:57 : 03/03
02:21:04 : 04/04
02:21:35 : 05/05
02:21:57 : 06/06
02:22:13 : 07/07
02:22:33 : 08/08
02:22:49 : 09/09
02:22:58 : 10/10
02:22:58 : Test finished.
----------
Total: 10/10
Probability that you were guessing: 0.1%
-- signature --
55a32e802b4cb013b10c6cff3bdd6353695015b0
Whereas with the original files with the test tones I could hear a difference at a reasonable playback volume which wasn't clipping the audible range's content [hard to say what's happening with the ultrasonics at that point, I don't hear them], not so with these new digital black files; I had to raise the gain 15dB more. It very likely is "clipping" level at this point, had there been actual content to clip, that is. However the noise I'm hearing now and keying on can't be deemed "clipping distortion" nor "IM distortion" since there is no actual signal.
[attachment=8370:2496_dig..._ds_1644.png]
I find it hard to believe I'm picking up on content that's 130 dB down [and at my age I conk out at 13k, maybe 14k tops]. How about a pic from your dual trace oscope of A's actual noise floor vs B's actual noise floor?
Update: I think with a quiet room I could even do the new digital black files at "0dB" volume setting on my preamp. I've had to overlook a faint hum problem (I would assume a ground loop) which I now realize I can kill by running my laptop on battery power instead of the AC cord. I'll see tomorrow if I can get the room quiet enough.
Update: I think with a quiet room I could even do the new digital black files at "0dB" volume setting on my preamp. I've had to overlook a faint hum problem (I would assume a ground loop) which I now realize I can kill by running my laptop on battery power instead of the AC cord. I'll see tomorrow if I can get the room quiet enough.
It is not news that in many systems the system gain can be artificially elevated to the point where normally inaudible noise can be reliably heard.
So, one key element of a reasonable test is using reasonable gain settings. Tests involving very low level signals such as these are especially error prone due to this influence.
I'm new to RMAA and can't figure out how to take snapshots from the analyzer part, but here's a screen grab with another program.
I see now your image wasn't theoretical but rather was an actual measurement. Oops. I'm still having problems believing this sort of thing would be audible to me at a playback level that although "loud" doesn't clip normal content, "0db" on my Audyssey calibrated prepro, but I guess I can.
I also can't remember how to bring my attachments I've successfully uploaded in another thread over to this one, but my RMAA analyzer screen grab shot [my very first one! Woo hoo] shows the same thing as your image.
Here's a link (http://www.hydrogenaud.io/forums/index.php?showtopic=109824&hl=) instead.
I didn't even download the samples but i see the pic. You still use some weird CEP settings for resampling?
I didn't even download the samples but i see the pic. You still use some weird CEP settings for resampling?
No weirdness, just careful tuning based on best knowledge I have about audibility. Q=150
You could lower yourself to try to ABX them for about 5 minutes... ;-)
Measurements related to AP24192 and Beach Audio Asynch USB
Beach Audio Xmos Asynch USB AK4490 DAC eBay (http://www.ebay.com/itm/321755642359?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT)
AP 24192 loop back
[attachment=8377:ap_24192_1_KHz.png]
AP24192 input terminated by 50 ohm
[attachment=8374:ap_24192__50_Zs_.png]
Beach Audio DAC 1 KHz with TRS->RCA adapters and RCA cables DAC Vout = 1.29 volts
[attachment=8375:ap_24192..._rca-trs.png]
Beach Audio DAC 1 KHz with Special TRS->RCA cables
[attachment=8376:ap_24192..._rca-trs.png]
Special RCA -> TRS cable
[attachment=8378:special_TRS_RCA.png]
44 KHz SR frequency response of Beach Audio Xmos Asynch USB AK4490 DAC (eBay):
[attachment=8379:ap_24192...dio_4424.png]
Notice roll off commencing @ 10 KHz and 5 dB down at 20 KHz.
Very similar to Pono player:
(http://1.bp.blogspot.com/-0M9Y7_CIDbg/VcLuJCQLGKI/AAAAAAAAFNQ/PMMyeKGDvqI/s1600/16-44+Spectrum.png)
Jitter test files - various degrees of audibility
No jitter (reference)
[attachment=8386:no__jitter.flac]
High, audible jitter
[attachment=8387:30_Hz_ma...tter_0.1.flac]
Harder to hear jitter
[attachment=8385:30_Hz_ji...vel_.025.flac]
[attachment=8411:MQA_Block_diagram.png]
Hardware ABX Comparator Schematics
[attachment=8412:ABX0001s.jpg]
[attachment=8413:ABX0002s.jpg]
[attachment=8414:ABX0003s.jpg]
[attachment=8415:ABX0004s.jpg]
Flac files of a piano clip from SQAM with various amounts of jitter added
Audio check 8 versus 16 bit resolution files as individual, downloadable flac files along with evidence of a reason why you should hear a difference even if you can't tell the difference between 8 and 16 bits (in context)
FFT of THD and noise from Behringer UCA test by NWAVGUY http://nwavguy.blogspot.com/2011/02/behringer-uca202-review.html
From Long-Term Durability of Pickup Diamonds and Records
Authors: Loescher, Friedrich A.; Hirsch, Frank H.
JAES Volume 22 Issue 10 pp. 800, 802, 804, 806; December 1974
Publication Date:December 1, 1974 Import into BibTeX
Permalink: http://www.aes.org/e-lib/browse.cfm?elib=2722
Bucket aligned multitones @ 2496 for technical tests
The test frequencies align with the bucket frequencies in a 96 kHz 65k point FFT so they tend to present more uniformly. In CEP these are about +/- 0.1 dB or better.
I threw in a swish for grins and giggles. This is a complex tone that will tend to present as a flat 65k FFT with a minimal amount of data.