If we look at the test signal he used, it's an asymmetrical waveform where half of it appears to be a standard sine wave, the other half has been "flattened" off which results in numerous harmonics if we are to display it on an FFT. One of the files is simply the inverted version of the other to test whether this polarity change is audible. Notice that the waveform is bandwidth limited, and there is no "ringing" after going through a digital filter.
Here are the waveforms overlaid on each other showing the fact that indeed they are just inverted versions of each other:
Since the signals are identical other than polarity inversion, the FFT shows that the signals contain the same frequencies whether in the "initial" orientation or inverted.
Beyond the digital waveforms on the computer audio editor, I can also confirm that playback of the waveforms on both the ASUS Xonar Essence One DAC and Oppo UDP-205 using bit-perfect WASAPI/ASIO were able to reproduce the waveforms accurately... Here's confirmation using a digital oscilloscope to examine the analogue headphone output of these DACs - basically I overlaid the tracing from the inverted and "normal" polarity waveforms to show that the DACs are reproducing the waves without gross asymmetrical distortion. Don't mind the temporal shift between screen captures from the devices:
Nice. We basically see that the headphone output from both machines do not show gross distortion whether the "flat" part of the waveform is positive or negative. I wanted to do this to make sure that it wasn't the DAC itself distorting and adding to audibility.
We can see that the ASUS Essence One's output is not as accurate compared to the Oppo - the channel balance isn't as precise with the left channel louder than the right. Notice how accurate the Oppo UDP-205 is! The 2 channels essentially overlap and unless you look closely, it's hard to tell that each tracing includes overlapped 2-channel stereo audio.
Here are the 3 headphones used, lined up for some rather stress-free (!) ABX listening:
My trusty old Sony MDR-V6 with velour replacement pads, the AKG Q701 "Quincy Jones", and the Sennheiser HD800 (I still love the "speed" and clarity of these things!).
As I posted last week, clearly the difference can be easily ABX'ed in foobar. ABX log for Sony V6 + ASUS Essence One:
ABX log for Sennheiser HD800 + ASUS Essence One:foo_abx 2.0.6c reportfoobar2000 v1.4.12019-06-15 21:10:58File A: AbsolutePolarityListeningTestTone.flacSHA1: c4f4e59eedb9e838e5c393c4508e1f29113591f0File B: AbsolutePolarityListeningTestToneInverted.flacSHA1: d093c9a9723042caf0b39d1afa8d814660100808Output:DSD : WASAPI (push) : Speakers (ASUS Xonar Essence One), 24-bitCrossfading: NO21:10:58 : Test started.21:11:10 : 01/0121:11:21 : 02/0221:11:31 : 03/0321:11:40 : 04/0421:11:50 : 05/0521:12:00 : 06/0621:12:09 : 07/0721:12:17 : 08/0821:12:25 : 09/0921:12:32 : 10/1021:12:43 : 11/1121:12:53 : 12/1221:13:01 : 13/1321:13:10 : 14/1421:13:20 : 15/1521:13:28 : 16/1621:13:28 : Test finished.----------Total: 16/16p-value: 0 (0%)-- signature --9bdb7ac25c2d9e1be17598806f81114426e5d48e
foo_abx 2.0.6c reportfoobar2000 v1.4.12019-06-15 21:28:19File A: AbsolutePolarityListeningTestTone.flacSHA1: c4f4e59eedb9e838e5c393c4508e1f29113591f0File B: AbsolutePolarityListeningTestToneInverted.flacSHA1: d093c9a9723042caf0b39d1afa8d814660100808Output:DSD : WASAPI (event) : Speakers (ASUS Xonar Essence One), 24-bitCrossfading: NO21:28:19 : Test started.21:28:29 : 01/0121:28:36 : 02/0221:28:42 : 03/0321:28:47 : 04/0421:28:53 : 05/0521:28:58 : 06/0621:29:03 : 07/0721:29:09 : 08/0821:29:15 : 09/0921:29:20 : 10/1021:29:25 : 11/1121:29:31 : 12/1221:29:37 : 13/1321:29:46 : 14/1421:29:51 : 15/1521:29:56 : 16/1621:29:56 : Test finished.----------Total: 16/16p-value: 0 (0%)-- signature --994dc2ea717ee73cd8b9ad7b05005def8970cc52
And here's the ABX log for the Sennheiser HD800 + Oppo UDP-205 (best combination of headphone and DAC output):
foo_abx 2.0.6c reportI even tested the AKG Q701 with my Microsoft Surface Pro 3 laptop headphone output from the laptop and the difference was clearly audible.
foobar2000 v1.4.4
2019-06-16 10:07:38
File A: 01 - Yamamoto - Absolute Polarity Listening Test Tone.flac
SHA1: 25f2888f75ec5bc87f7c26058912756a8c353682
File B: 02 - Yamamoto - Absolute Polarity Listening Test Tone Inverted.flac
SHA1: 03837b9de806fdbe16cb661446bfbedb9540c985
Output:
ASIO : OPPO USB AUDIO 2.0 ASIO Driver
Crossfading: NO
10:07:38 : Test started.
10:07:47 : 01/01
10:08:01 : 02/02
10:08:10 : 03/03
10:08:17 : 04/04
10:08:25 : 05/05
10:08:34 : 06/06
10:08:43 : 07/07
10:08:53 : 08/08
10:09:02 : 09/09
10:09:11 : 10/10
10:09:19 : 11/11
10:09:28 : 12/12
10:09:37 : 13/13
10:09:49 : 14/14
10:09:58 : 15/15
10:10:07 : 16/16
10:10:07 : Test finished.
----------
Total: 16/16
p-value: 0 (0%)
-- signature --
112f875b13d1c7320f93addd419d4e3258d92eb1
Initially I did wonder whether the polarity difference was more audible with the ASUS Essence One versus the Oppo UDP-205. Hard to say once I started running the ABX test. Either way, like Yamamoto-san, I was able to achieve 16/16 correct. I got 15/16 on the AKG Q701 + Surface Pro 3 combination so maybe that's indication that a lower fidelity device might make it less easy to hear.
With each DAC and headphone the subjective sound was indeed consistent and similar to Yamamoto's description on his page. While I did not perceive any right/left change in balance (other than that R-L imbalance shown above with the ASUS Essence One), I agree that the non-inverted version sounded "thicker", more "direct". In comparison the inverted sample seemed to be a little "thinner", tonally slightly higher pitched to my ears, perhaps a little more "laid back".
While it was easier with headphones, it was also not a problem with my AudioEngine A2 desktop speakers - 15/16 (p<0.0003) correct match using ABX Comparator.
For completeness, since my family was asking what the heck I was doing with this annoying sound, I gave the ABX test to my wife and kids through the Sennheiser HD800 and ASUS DAC :-). My wife and I are both in our mid-40's, my son is 14 and daughter 12 years old. No problem with ABX score of at least 14/16 correct for everyone (easily p<0.05).
Therefore, my expectation is that Yamamoto's test files should be generally audible. I would definitely expect "golden ears" to have no problem with hearing the polarity effect with this test file.
Conclusion...
Well guys and gals, indeed absolute polarity/phase is an audible phenomenon. This is an interesting property of the physiology of human hearing. I'm a believer! And was definitely wrong last week to assume that "absolute polarity" was about whether the first wave was up/down going, and the J.G. Holt Stereophile article from 1980 could use a little updating to include the physiology of the human inner ear.I looked into what we understand about the mammalian hearing mechanism, and indeed, it looks like the mechanicoelectrical sound transduction with release of neurotransmitters from the cochlear hair cells is directional in nature (much of the basic research appears to have appeared decades ago, here's a 2017 review paper from Comprehensive Physiology addressing some of the known details these days). "Positive deviation" of the hair cells will open primarily potassium and calcium channels whereas negative deviation closes these channels.
This results in a "half-wave rectification" effect in our hearing mechanism... Polarity of the basilar membrane deviation in the cochlea will move the hair cells, trigger action potentials that induces neurotransmission (glutamate release and stimulation of post-synaptic AMPA receptors) in one direction. Check out this detailed PowerPoint presentation which looks like the slide set from University College London, especially slide 42 regarding the rectification process.
This past week, I also had the pleasure of conversing with Matthias Carstens of RME about this. He posted some test files using sawtooth waves to demonstrate the audibility of absolute phase as well in 2017. Here's the waveform from the 200Hz test file at the point of the polarity flip:
Have a listen. Again, the difference in sound is very evident. Matthias notes that the effect is reduced with higher frequencies and almost gone by 1kHz.
In the comments section last week, Yamamoto and GillesP discussed how trombone music tends to show waveform asymmetry. Brass instruments like trumpets and trombones tend to show waveform asymmetry because the air is blown forward and there is impedance of vibrations back towards the mouthpiece. I imagine that this would be especially pronounced with close miking.
Here is a file package you can download - have a listen to this. It contains both the trombone sample track from GillesP last week (correct labeling for the inverted track), plus a 26-second segment from Trombone Shorty's "Laveau Dirge No. 1" from his album Parking Lot Symphony (2017), a more complex piece, with backing instruments. As usual, if I post audio files containing copyrighted content, the sample is provided on the basis of Fair Use for the purpose of education and commentary. Please delete when done and purchase the music if you like it.
This is what the waveforms look like with the asymmetry evident; "normal" polarity versions shown:
 |
| The short segment from Gilles. |
Can you hear a difference between the "normal" polarity and the inverted polarity version of each of the tracks? Does one version sound "fuller", more "real"? Are there differences in the perceived soundstage between normal polarity and when inverted?
Personally, while I can easily pass the ABX test with the synthetic samples from Yamamoto and the difference is obvious with RME's sawtooth files, I do not hear enough of a difference with Gilles' example or this Trombone Shorty segment to be able to pass a blind test. For me, it seems that although the effect can be demonstrated with test tones, the effect disappears with more complex audio, it appears to be unnoticeable (to me at least). Let me know if you can ABX a difference!
To be clear, remember that what I said last week could still be true about asymmetrical distortions from transducers. Do not assume that just hearing a difference automatically means that one's speakers, amps, or DAC is "good" - double check that you're not hearing a system defect first!
Ultimately, as a "perfectionist" audiophile, intellectually, I think it would be elegant to maintain absolute polarity from the studio all the way to home playback to maintain the highest fidelity. However, I'm not sure if this would improve the sound, nor provide greater enjoyment. Furthermore, other than the most direct 2-channel recordings of acoustic performances, I don't see how multitrack studio productions can ensure that every microphone input, every (pre)amp, every sample, of every track be in "absolute" polarity. The cost in time and effort just doesn't seem beneficial; perhaps even impossible in some situations.
Maybe this is a good example of one of those times where we as audiophiles might say "everything matters" (including the physiology of hearing absolute polarity/phase with isolated test samples), but compared to a multitude of other factors that make sound/music "good", it's probably reasonable to think of this as being much lower on the list of priorities.
After I wrote that conclusion above, I read this comment from reader mp which I think sums things up very nicely:
Folks interested in the audibility of 'absolute phase' are referred to Audio Critic issues 18 and 22 letters section (on-line with a search). Another commentor cites it, but I'd like to expand. In the early '90s a man named Clark Johnsen was citing something called the "Wood Effect", arguing that absolute phase was not only audible, but a strong determinant in reproduced sound quality. He wrote a book about it, I think. Of course he did not 'prove' his claim using DBT, but made an assertion, and came up with a 'new' test protocol that, in his opinion, could be used to demonstrate phase audibility to anyone's satisfaction.
Jeff Corey (professor of experimental psychology) critiqued Johnsen's "triple blind test protocol" as nonsense, and R.A. Greiner (professor emeritus EE) wrote in to discuss his recent JAES paper on polarity. Dr. Greiner's experiments showed:
1) acoustic polarity is audible with steady state (monotone) signals.
2) polarity with simple musical passages was 'extremely difficult' to ascertain in a 'highly idealized simplified laboratory environment'.
3) on complex musical passages in a normal living room no subject was able to reliably determine phase differences.
Dr Greiner stated that he discussed the issue with both Drs (Richard C) Heyser and (Stanley) Lipshitz at AES meetings. Their joint conclusion was that while it would be ideal if the recording industry adhered to a standard, in the scheme of recording problems absolute phase is pretty low, to not a concern at all, in the list of special effects that need addressing.What a perfect summary with the 3 points, thanks mp! I see that not long ago, there's this Positive Feedback article from Johnsen, so it looks like the discussion continued. Notice in that article the allegations of "certain agencies" being aware of this polarity issue and the alleged "subterfuge" - I don't believe this is healthy thinking.
One last thought before signing off - isn't modern audiophilia great!? Obviously, "absolute phase" is a topic much discussed over the decades (Greiner referred to his presentations and papers starting in 1991 and his letter in Audio Critic issue 22 from 1995 is great). However, there is a huge difference between reading that Issue 22 letter, imagining what Greiner did in his "Electroacoustics Laboratory" if we were in the 90's versus me downloading these test files, examining the contents in an audio editor, putting the signal through an inexpensive oscilloscope and ultimately listening to them myself in 2019 to develop a sense of the experience of the perception. That link between subjective experience and objective verification is something that all audiophiles can partake in if we wanted to.
Isn't it wonderful that these days we have the technologies easily available for transfer of ideas, images, and of course music files to experience first hand?! Isn't it wonderful that highly accurate consumer gear is so easily within reach and testing equipment is essentially ubiquitous to anyone who wants to explore and understand for ourselves without needing to go through intermediaries (especially the "high priest" audio reviewers telling us what to believe!)? Of course, this does not mean we're all "experts", but with experience and discussion with others, we can all learn and up our game in accruing knowledge and wisdom beyond what we're simply told...
While there are numerous concerns we can raise about the Internet like all the "fake news" out there, pseudointellectual nonsense, antisocial fringe groups, concerns around privacy, active monitoring by organizations, etc. let's not forget the benefits of this freedom and the innovations these technologies catalyze. This combination of freedom of access to knowledge, sharing of ideas, and the means to seek truth through empirical techniques IMO represents what I believe is a much necessary rise of a rational and "more objective" audiophile hobby. One where the younger generations of audiophiles need not conform to the "received wisdom" of nonsense like $$$$ cables having major effects on sound, ridiculous jitter FUD, or bowing down to unnecessary "new formats" designed for corporations and promoted by a sponsored press.
While I have been critical of many aspects of audiophilia, as an audiophile, I am personally very excited about the future of this hobby! Perhaps some of what we discuss here and what I see around the Internet with the sharing on forums is representative of what the future holds. A democratization of knowledge, encouragement for experimentation, and openness to mindfully experience the truth around music reproduction technology. While we can never anticipate the irrationality of humans, I believe that the rise of more reasonable and knowledgeable voices discussing audio hardware will result in rewarding the true commercial innovators while the fate of snake-oil artists by right eventually results in obscurity and ridicule. Such is necessary for the integrity and health of any industry (and IMO there are many industries these days that must face this beyond audio).
Have a great week everyone! Happy summer and I hope you're enjoy the music...
Thanks for your article! Your subjective sound impression is interesting. My right ear had suffered middle ear otitis media in early age and its sound quality is somewhat lesser than the left ear. This may cause left/right drift of sound image. In fact I feel this right/left sound drift is the most obvious difference among all sound characteristics differences of absolute polarity :)
ReplyDeleteWhen creating this test signal, I listened to several different variations and found there is the best height of plateau :) Also sound difference becomes much smaller when the signal frequency is one octave higher than this.
Matthias's test files are interesting, 20Hz sound is very similar to pedal tone of Tuba. 100Hz Sawtooth is the most obvious different one for my ears.
BTW I found the absolute polarity sound difference with the following music, sound is slightely more bright when polarity is inverted: https://youtu.be/HIAKFkvKXYY?t=687
Found another music example, this one has more obvious sound difference https://youtu.be/2lyg809cbWw?t=28
DeleteInteresting about the right-left ear difference Yamamoto.
DeleteIt's interesting that sometimes our own biological "imperfections" actually can accentuate anomalies that perhaps we might not notice as much of otherwise! Years ago with the MP3 test, I know that some people with hearing anomalies were better able to pick up the difference between MP3 320kbps vs. lossless better than "normal" ears...
I'll have to check out those example. Wow, video game synthetic music!
Hi Archimago,
DeleteIn case you don't know, that music from the game "Cheetahmen" Yamamoto-san mentioned is a masterpiece. I also made a remix 10 years ago.
https://www.nicovideo.jp/watch/sm6771142
You may need to stream the video above at non busy hours (2am - 6pm Japan Time?) to ensure high quality streaming.
Very interesting! I guess it’s normal that ears have a cellular directional mechanism since we locate sound in space by phase differences so there must be some absolute mechanism in each ear for comparisons to work.
ReplyDeleteI was pretty sure of my short example’s initial preference for correct polarity even before I looked at the waveform but to be sure I did manage to run a foobar abx test finally. I’m on a Mac and I had to install it in a virtual machine running Windows Xp with VmWare since the Mac version doesn’t support components.
Pretty good results using only a very short portion at the beginning with AKG K702 headphones plugged into my MacPro:
—
foo_abx 2.0.6c report
foobar2000 v1.4.5
2019-06-23 10:03:19
File A: trombone_inv.flac
SHA1: aefa903b638ff64fb17ec0e2bbcd257d7f70d55f
File B: trombone.flac
SHA1: e5a919a1ae1fa3af35f8468099f467253b6a91f0
Output:
DS : PÈriphÈrique audio principal, 16-bit
Crossfading: NO
10:03:19 : Test started.
10:04:02 : 01/01
10:04:15 : 01/02
10:04:25 : 01/03
10:04:37 : 02/04
10:04:47 : 03/05
10:05:01 : 04/06
10:05:15 : 05/07
10:05:25 : 06/08
10:05:45 : 07/09
10:06:05 : 07/10
10:06:33 : 08/11
10:06:49 : 09/12
10:06:59 : 09/13
10:07:15 : 10/14
10:07:25 : 11/15
10:07:45 : 12/16
10:07:45 : Test finished.
----------
Total: 12/16
p-value: 0.0384 (3.84%)
-- signature --
d44f0fa522392dc4c7b2027d36e45e1262c6a2be
Nice work Gilles! Impressive as ever :-).
DeleteSince I didn't find the trombone sample as easy, your result has inspired me to devote a quiet evening to the ABX testing after the kids go down.
At least with polarity, one can sort of argue that getting it right in the studio possibly, just possibly, and under the right conditions, very limited conditions, makes things sound better. Or at least different. Which is more than you can say for most of the tweako stuff that is sold, or propagated in audiophile-land.
ReplyDeleteThe problem as I see it is when people like Clark Johnson claim that it is a very important issue, and that getting in right will change your musical life for the better.
Here is a quote from Clark that I tracked down:
"Masked by random combination with other distortions in the music reproduction chain, an unsuspected major contributor has lain hidden: Aural sensitivity to ‘phase inversion’ — the Wood Effect."
As anyone who has bothered to check it out quickly finds, it's not a 'major' distortion factor within the recording chain. Never has been, and never will be. And 'aural sensitivity' to polarity quickly fades once one starts listening to real music. On the other hand, if your preamp is a Krohn Hite oscillator, and test tones are your musical bag, then I suppose Johnson has a point.
Yup mp. A highly important point and one we've seen over and over again in the audio press and among folks like Johnsen with a thesis to "sell".
DeleteThis all reminds me of jitter and how based on some people (like ahem... Ted Smith and the purveyors of USB "clean up" accessories), it too is a "major contributor" to sound. In fact, I would contend that "absolute polarity" is more audible than jitter ever was with reputable equipment.
Can't help but think that this is what happens when the Industry is trying to sell products with mature technology. It's necessary to find explanations as to why something is better than another and often actual differences become overly magnified.
One more addition if I may. Greiner wrote an article about this topic in the 12/93 Audio Magazine, which can be found on line at americanradiohistory dot com archives. In an on-line forum, Clark Johnson typed that 'he had heard from a third party' that Greiner's negative findings were influenced by speakers that were not 'resolving.' This sort of critique is often heard by subjectivists, but they usually don't offer criteria about how and why a particular speaker is not 'resolving enough.' In any case, Greiner's findings were positive, just not in the way Clark Johnson would have liked them to be. That is, a big problem.
ReplyDeleteGreiner stated that, in fact, polarity on simple signals was easily heard on his test speaker set up with simple tones. Interestingly enough, no one was able to hear it on music within a two speaker stereo array. Therefore, his experiments were conducted using a single quad amplified speaker, in a 20x20 'nearly anechoic' (above 250Hz) room. The system's FR was said to be uniform at the listening level, with low distortion at moderate levels. Headphones were also used. When polarity was reliably demonstrated, it could be heard with both the speaker and headphone.
He discusses how the real problem with polarity occurs in a recording studio, beginning with microphones, mic preamps, mixing boards, recorders, etc. Then mastering. With today's multi-track recording you could easily have one instrument recorded in phase, and another with reversed polarity. How would one even begin to tell who got it right? A very minimalist recording setup would be the only way to make sure.
Nice details mp...
DeleteYeah, begin with the old "not resolving enough gear" complaint for why a result isn't as expected. If that fails, then go to the next level and use the "ears not golden enough" explanation :-).
I agree the more complex the music the harder it is to hear the effect. And of course global polarity in multi-track recording is unreliable. Also some playing techniques like for the oboe where it is common to output sound while breathing out and in successively make the polarity unimportant. In any case this is something not primordial in overall audio quality. I did manage to find a section of the more complex Trombone Shorty segment where I could distinguish with abx testing with foobar. First try with concentrating on attacks at the beginning did not work well, but on second try with the entry of many instruments at about 00.17 I got quite good results:
ReplyDeletefoo_abx 2.0.6c report
foobar2000 v1.4.5
2019-06-23 17:13:10
File A: Trombone Shorty - Laveau Dirge No. 1 [Normal Polarity].flac
SHA1: b53e79740c1555f552ed7ea76e41b6f0232c4683
File B: Trombone Shorty - Laveau Dirge No. 1 [Inverted Polarity].flac
SHA1: 882a77bf606641cc728e823f0b6a38bf42cb0921
Output:
DS : PÈriphÈrique audio principal, 16-bit
Crossfading: NO
17:13:10 : Test started.
17:17:39 : 01/01
17:17:56 : 01/02
17:18:16 : 02/03
17:18:27 : 03/04
17:18:39 : 04/05
17:18:51 : 04/06
17:19:00 : 05/07
17:19:14 : 06/08
17:19:30 : 07/09
17:19:39 : 08/10
17:19:54 : 08/11
17:20:08 : 09/12
17:20:31 : 10/13
17:20:40 : 11/14
17:20:47 : 12/15
17:20:56 : 13/16
17:20:56 : Test finished.
----------
Total: 13/16
p-value: 0.0106 (1.06%)
-- signature --
5a5a9d20426570dcbdeaf53204b04cb441c917b6
Wow!
DeleteVery impressed Gilles. You're certainly the most diligent and capable listener I know. Enjoy the music with those golden ears, man!
Hi Archi.
ReplyDelete2 great articles about absolute polarity. Yes, this is audible especially with small ensembles or solo instruments
and as you have mentioned, has to do with the hair cells in the ear, together with the unsymmetrical band pass filters followed by halve wave rectifier
that is also responsible, the an upward tone sweep does sound different than the same sweep downward,
that you also can see, when you doing a simulation of these ear bands and rectifiers, as it has been done at Fraunhofer, when developing the MP3 coder.
Surprisingly, that this does found less interest in the press than some voodoo stuff, as does the phase behavior of loudspeakers.
Thumbs up.
Juergen
Thanks Juergen. Appreciate your mention of the development of the MP3 coder as well. I imagine that this effect must be something the psychoacoustic modeling systems must be interested in to maintain perceptual quality.
DeleteI do find it interesting that typical audiophile magazines don't talk about "real" effects like this as well... I guess that's why it took me till now to even look into this since I rarely read/hear about this effect (maybe every few years run across an article or forum comment).
I don't think it's a stretch to suggest that the audiophile press is not talking about this because there's not much they could sell based on this other than to mention when an amp may have a phase inversion switch or setting I suppose. Now that this can be easily done in software playback, it's probably not even much of a feature that many audiophiles would even talk of.
To me this is rather unfortunate of course... It's like the whole issue of severe dynamic range compression; an obviously very real issue affecting so much music these days! Yet barely a peep, minimal discussion to raise awareness among audiophiles, or any effort to represent audiophiles who are interested in high quality, high fidelity recordings.
Here's an idea Audiophile Press. When you're talking to representatives from Warner, UMG, and Sony (about the disaster called MQA for example), mind putting a good word in for something that actually makes a positive difference like asking if they could fix all those horrendous DR5 albums?!
Before simply claiming that "hi-res" is all that great, Mr. Audiophile Magazine Editor, mind taking a moment to actually look if there is any content that needed bigger storage and higher streaming bitrates in some of your recommended recordings? Yeah, I know you and your reviewers all "hear" the better sound quality (of course, since you're all golden ears), but it wouldn't hurt to check things out once awhile and suggest to your readers how to separate the wheat from the chaff, right?
An actual "reality based audiophile press" is what we need. I personally won't be holding my breath waiting on this though. :-(
foo_abx 2.0.6c report
ReplyDeletefoobar2000 v1.4.5
2019-06-26 10:00:46
File A: Trombone from Gilles [Inverted Polarity].flac
SHA1: cb9d20486748c6fcd56b8d949515497e3aff0ca8
File B: Trombone from Gilles [Normal Polarity].flac
SHA1: 950d5424f3741fdb677c9a2790cf72a63118f91b
Output:
DS : Primary Sound Driver
Crossfading: NO
10:00:46 : Test started.
10:03:49 : 01/01
10:09:05 : 02/02
10:09:16 : 03/03
10:09:36 : 04/04
10:09:54 : 05/05
10:10:11 : 06/06
10:10:47 : 07/07
10:11:02 : 08/08
10:11:24 : 08/09
10:11:45 : 09/10
10:11:56 : 10/11
10:12:04 : 11/12
10:12:12 : 12/13
10:12:40 : 13/14
10:12:53 : 14/15
10:13:02 : 15/16
10:13:02 : Test finished.
I am using a Benchmark DAC2 feeding a Benchmark HPA4 headphone amplifier driving a pair of HD650 headphones.
I am listening to the fade of the last note and comparing the prominence of the overtones. Curiously, I find the overtones more pronounced on the inverted version than on the non-inverted. This is the opposite of what I observed with the sinusoidal tones. The sinusoidal tones were extremely easy to distinguish even when listening through my Benchmark SMS1 loudspeakers driven by a Benchmark AHB2 power amplifier.
One other observation:
ReplyDeleteIt was easier to hear the effects of inversion when using the polarity switch on the front of the DAC2. This switch is instantaneous, but the foobar ABX produces a brief mute whenever sources are switched. The brief mute makes the comparison harder. To mitigate this, I unchecked the "Keep playback position when changing track" button. This way I was always comparing to the same starting point.
Prominent overtones to subdued overtones sounds like a slight downward pitch shift. Subdued overtones to prominent overtones sounds like a slight upward pitch shift.
On each trial I played XY and then YX to see which one produced an apparent downward pitch shift. This is somewhat complicated by the fact that the fading tone is dropping in pitch, but the effect is audible (as confirmed by the test results above).
Interesting observation. In my case I concentrated more on the first note attack that seems to follow the visual curve of the envelope. If you concentrate, the correct polarity note sounds more like wwaaaaauuu and the other like wwuuuuuaaa...
DeleteOf course, listening casually both sound similar so that musically it is not really concerning.
This reminds me of optical illusions like the Necker cube where looking longer than necessary brings out another solution to the ambiguity.
Interesting. The aaauuu and uuuaaa that I heard coincided with the switching between the two sources. The inverted produces an aaa on steady notes while the non-inverted produces an uuu on steady notes (relative to the inverted). The difference between aaa and uuu seems to be the audibility of the overtones where aaa corresponds to a greater audibility of the overtones. The pronounced overtones make the aaa sound slightly higher in pitch than the uuu although there is no pitch difference.
DeleteStrangely, the clipped sine wave test gave me the opposite impression. The overtones were more pronounced on the non-inverted version of the clipped sine waves.
Output:
DeleteDS : Primary Sound Driver
Crossfading: NO
16:12:29 : Test started.
16:12:43 : 01/01
16:12:48 : 02/02
16:12:54 : 03/03
16:13:00 : 04/04
16:13:04 : 05/05
16:13:08 : 06/06
16:13:12 : 07/07
16:13:19 : 08/08
16:13:22 : 09/09
16:13:26 : 10/10
16:13:32 : 11/11
16:13:37 : 12/12
16:13:42 : 13/13
16:13:47 : 14/14
16:13:53 : 15/15
16:14:02 : 16/16
16:14:02 : Test finished.
----------
Total: 16/16
p-value: 0 (0%)
-- signature --
1337f615f97fad812febb453d1b0a59c93e9d42a
These are my ABX results for the clipped sine wave polarity test. These results demonstrate how audible this effect is with this test tone. It only took me 90 seconds to complete 16 trials with 100% accuracy.
DAC2 > HPA4 > HD650
Nice one Gearbuilder!
DeleteGreat gear as well you're using to do these listening tests.
foo_abx 2.0.6c report
ReplyDeletefoobar2000 v1.4.5
2019-06-27 16:12:29
File A: AbsolutePolarityListeningTestTone.flac
SHA1: c4f4e59eedb9e838e5c393c4508e1f29113591f0
File B: AbsolutePolarityListeningTestToneInverted.flac
SHA1: d093c9a9723042caf0b39d1afa8d814660100808
Output:
DS : Primary Sound Driver
Crossfading: NO
16:12:29 : Test started.
16:12:43 : 01/01
16:12:48 : 02/02
16:12:54 : 03/03
16:13:00 : 04/04
16:13:04 : 05/05
16:13:08 : 06/06
16:13:12 : 07/07
16:13:19 : 08/08
16:13:22 : 09/09
16:13:26 : 10/10
16:13:32 : 11/11
16:13:37 : 12/12
16:13:42 : 13/13
16:13:47 : 14/14
16:13:53 : 15/15
16:14:02 : 16/16
16:14:02 : Test finished.
----------
Total: 16/16
p-value: 0 (0%)
-- signature --
1337f615f97fad812febb453d1b0a59c93e9d42a
Thank you for highlighting this very interesting phenomenon. The difference between the two of Yamamoto's waveforms is so obvious it's not even funny. No golden ears required. Do I understand correctly that it happens because our hearing has different transfer functions for the positive and negative pressure changes?
ReplyDeleteI am able to successfully blind test inverted polarity on many of my songs. What I initially thought was intermodulation distortion on my Moondrop Variations at 1:14 in Il Vento D'oro from the JOJO's Bizarre Adventure - Golden Wind Soundtrack when he says "golden," is actually caused by the inverted absolute polarity of the Variations. I sold them over this and would have sent them out if I did not realize the "warbling" was caused by the inverted absolute polarity. Inverted polarity absolutely sounds lower fidelity here and I have successfully blind ABX'd the inverted vs correct file to a 99.99% confidence interval.
ReplyDeleteIn addition, in Signs of Love from the Persona 4 soundtrack, a clear difference is audible in the initial transient with inverted absolute polarity adding artificial bite and damaging timbre.