Blue or red pill?

[PeterSt]

Of course it is not by 100% guarantee that you saw the same thing, Paul, but I would put a lot of money on it.

 

Also, from the start you have bitten yourself into this (together with @STC btw) so if anyone could do it, it was you. What is also important IMO is that you "saw it happening" hence were motivated to find something until dropped dead (here, with due respect STC has been slightly different although with the very same good cause).

 

What you should do now (for yourself and only if you like it) is highlight these differences by means of showing the waves themselves. OK, you just did, but you have to be able to show it by means of the two separately instead of the more abstract diff. Understand ?

Only then it will be visualized what actually creates this show with one variable only : you will not know the distance of the two time cursors (so to speak - I hope I am clear a little). This, while my test as of then just had one time cursor in the one situation (recording) and no time cursor in the other.

 

Edit : This is post 2500. This means that it took 2498 posts to get there. Phew. 

[PeterSt]

1 hour ago, pkane2001 said:

 

With imagination you'd even see the two cursors against each other; if you zoom in horizontally a bit more this will be more clear, I think.

[STC]

1 hour ago, PeterSt said:

Also, from the start you have bitten yourself into this (together with @STC btw) so if anyone could do it, it was you. What is also important IMO is that you "saw it happening" hence were motivated to find something until dropped dead (here, with due respect STC has been slightly different although with the very same good cause).

 

I am not sure if any of contribution here is relevant but I was motivated to find the reason for another selfish reason of mine. In short, I suspect the DAC or PC do not behave similarly under different load. In my case, whenever I attempt to due quick comparison with my system one DAC timing shifts by 12ms. It will remain constant for months and in sync but the moment I do quick AB of different combination of other parameters the 12ms out of sync sets in.

 

 

IMO, In this blindtest, Foobar ABX with headphones is not going to show enough difference for you achieve the accepted level of confidence. The two analogue files exhibit some differences. The difference can affect the perceived sound over loudspeakers sitting at a distance as the loudness level drops by 6 dB for doubling of the distance.

 

Let's say there is an image created by the interaural level difference between the left and right channel of just by 3.5 dB. Let's say the reference level is at 46.5dB for the left and the right channel is 50 dB (say the measurement is taken at one meter). At your sitting position of 4 meter the left speaker level would have dropped to 34.5dB and the right channel will be 38dB. Assuming your noise floor of the room is 35dB, the image would be in a slight different position compared to a slight difference of the other track of 47dB and 49.8dB. You have to also consider the phase of both channels. Any slight difference there can be amplified with distance more than headphones do.

 

The question is to ask whether the difference can be identical at all time? Can you hold your head steady in the exact spot for so many different ABX trial? There can be also slight difference in the radiation pattern of the different loudspeakers than can amplify the difference. And the consistency of the software performance. Static built up, unclear cache and etc etc can affect the sound.

 

Another thing to consider is - how consistent is the DAC output under various quick switching? In my experience, when I change tracks very quickly some of the DAC timing goes off track slightly. Can such anomaly exist during quick switching with ABX? BTW, I can tell you that not many actually can hear the 12ms difference when I do the demo.  

 

BTW, try using Musicscope to analyze the difference. It shows some difference especially in the StereoMeter that can explain the shift in image.

[esldude]

4 hours ago, pkane2001 said:

 

Have you guys tried to zoom in on the delta waveform after subtracting A and B? Here's what I get:

 

This appears to have a beat at about every millisecond, or 1000Hz. Is that the quantization error we were looking at before? Shouldn't it get zeroed-out when subtracting A and B? Or is the quantization error different between captures due to noise?

 

Curious if you get the same, as I am still not sure that diffmaker is working properly, so can't be sure of my results.

I'm not seeing that.  Is this 3a minus 4b and trimmed to equal length with sample rate correction?

[PeterSt]

2 hours ago, STC said:

In short, I suspect the DAC or PC do not behave similarly under different load. In my case, whenever I attempt to due quick comparison with my system one DAC timing shifts by 12ms.

 

Can you briefly tell what actually shifts by 12ms and how you can tell it indeed is 12ms ? You say "one DAC" but it could be that your meant to say "one channel".

 

When you started this subject something in this thread I already wanted to respond to it but there too I did not quite understand ...

[PeterSt]

11 minutes ago, PeterSt said:

You say "one DAC"

 

OK, I think I can see now that this is about your ambiophonics setup and syncing. Of course that gives me no clue really of what happens with the 12ms (but I think it is important and that is why I ask).

[STC]

36 minutes ago, PeterSt said:

 

Can you briefly tell what actually shifts by 12ms and how you can tell it indeed is 12ms ? You say "one DAC" but it could be that your meant to say "one channel".

 

When you started this subject something in this thread I already wanted to respond to it but there too I did not quite understand ...

 

It is hard to put across my point due to my limited understanding in this subject. Some of the terminology may not even mean what I intended to say....anyway...

 

Not channel. I use multiple DACs and used REW measurements to synchronize all speakers to perfectly align so that the signals from the speakers arrive exactly at the same time to the microphone at the listeners location. The actual time difference shown in the REW and the value shown in the slider varies but that is another matter.

 

For an example, the signal from PC split to two DAC.  The DACs are of different make and therefore there will be inherent time difference. This difference can be adjusted using the delay feature. The perfect sync happens when one DAC's input signal is delayed by 55ms. However, when I do quick changes for AB'ing to show difference between two different tracks, quiet often the delay becomes shorter and the new value is 43ms. The changes happen only when I intervene to do the switching. In normal operation, it can play even 100 songs playlist from different album and drive without missing a beat.

 

I am not sure if it applies to other software. I am using Jriver and zoning. 

 

edit: 55ms is only applicable to one DAC. Each DAC carries a different delay ranging from 40 to 86ms in reference to the main DAC.

 

23 minutes ago, PeterSt said:

OK, I think I can see now that this is about your ambiophonics setup and syncing. Of course that gives me no clue really of what happens with the 12ms (but I think it is important and that is why I ask).

 

It is not related to ambiophonics. It is playing other speakers with convolution. No ambiophonics there. 

 

[fas42]

20 hours ago, mansr said:

I upsampled the 6 analogue captures of the music track 64x (to ~11MHz) and aligned them. Then I plotted the power spectra for the differences between each AA and BB pair, three of each.

 

 

 

I'm not following these plots - are you saying you achieved nulls of 140dB down between the versions when sufficiently aligned?

[fas42]

9 hours ago, psjug said:

What do you mean by glitch?  Something else besides the ultrasonic stuff?

 

Most of the ultrasonic stuff - the obvious noisy stuff was filtered out, but there was residual, just above 20kHz murmuring going on ... was this significant, was it an artifact of the recording process?

[pkane2001]

5 hours ago, esldude said:

I'm not seeing that.  Is this 3a minus 4b and trimmed to equal length with sample rate correction?

 

Yes, exactly.

 

Out of curiosity, what version of diffmaker are you using? Mine is 3.22.

[pkane2001]

9 hours ago, PeterSt said:

Paul, you have caught it.

Very very well done.

 

I’d still like to figure out if the audio diffmaker is the cause of what I’m seeing before accepting any of this is real

[psjug]

On 4/17/2018 at 12:39 PM, pkane2001 said:

Note the absence of the extra frequencies that were there when examining A3-B4 residuals (posted earlier):

 

@pkane2001  I think I might know why you are getting these peaks in your spectrum from the diffmaker difference file.  When you do the FFT you need to make sure you don't include the very beginning, because of the alignment adjustment.  I noticed that when I forgot to cut this part out that I got these same spikes.

[psjug]

5 hours ago, fas42 said:

 

Most of the ultrasonic stuff - the obvious noisy stuff was filtered out, but there was residual, just above 20kHz murmuring going on ... was this significant, was it an artifact of the recording process?

If there is any problem in that band it seems to be the same on A and B.  Diffmaker is only showing file differences in the higher ultrasonics, where random noise does not subtract out.

[pkane2001]

12 minutes ago, psjug said:

@pkane2001  I think I might know why you are getting these peaks in your spectrum from the diffmaker difference file.  When you do the FFT you need to make sure you don't include the very beginning, because of the alignment adjustment.  I noticed that when I forgot to cut this part out that I got these same spikes.

 

Are you applying some sort of window at the beginning and the end, or just cutting off a second or two? The delta file I posted was generated after removing 1 second off the beginning of both A and B and removing a few seconds at the end to ensure they at least looked the same when zoomed in on both ends. Do you normalize the files before feeding them to diffmaker? I didn't.

[psjug]

Just now, pkane2001 said:

 

Are you applying some sort of window at the beginning and the end, or just cutting off a second or two? The delta file I posted was generated after removing 1 second off the beginning of both A and B and removing a few seconds at the end to ensure they at least looked the same when zoomed in on both ends.

I'm not sure we are following each other or not.  I trimmed the ends off both input files so they were the same same length.  Then after the difference file is generated, there is a blip at the beginning because of the time adjustment.  So before running a FFT on the difference, select only the good part of the difference waveform, not including the very beginning (or very end?).

[pkane2001]

1 minute ago, psjug said:

I'm not sure we are following each other or not.  I trimmed the ends off both input files so they were the same same length.  Then after the difference file is generated, there is a blip at the beginning because of the time adjustment.  So before running a FFT on the difference, select only the good part of the difference waveform, not including the very beginning (or very end?).

 

That's pretty much what I've been doing when posting spectrum plots (selecting a portion without the ends). I'm not sure why, but it looks like diffmaker is working differently for me than for you or for @esldude. I'll hold off on claiming any results until I can get it to function in a similar way. Are you using it on Windows?

[psjug]

11 minutes ago, pkane2001 said:

 Are you using it on Windows?

I could not get diffmaker to run on Windows 7 or Windows 10.  I've been running it in virtual XP.  I'm using version 3.22 same as you.

[psjug]

On 4/17/2018 at 3:23 PM, esldude said:

I did the same thing brickwalling at 30 khz to -120 db at 40 khz.  Why?  Because the ultrasonic noise is louder than the quieter parts of the music.  It becomes the bulk of the signal at times.  When I did this and repeated Diffmaker it made little to no difference.  Not even a db difference in the null depth for the most part.  So that is simply residual noise from noise shaping and not effecting the musical signal which was originally from 44.1 sources. 

It's not really important, but I have been wondering about this.  I assumed Diffmaker's null depth number is an average across the spectrum.  And so the noise in the higher ultrasonics in these captures would skew the report.  But you say you got the same null depth number filtered and unfiltered.  So then how IS Diffmaker's null depth number calculated?

Edit: never mind... I see there is a "null depth evaluation range" setting.  So that explains it.

[esldude]

I think we are cluttering up the thread talking about Diffmaker's inner workings.  I know we are talking about results with the files used.  Maybe a separate thread about the inner workings, pitfalls, and how much you can trust Diffmaker would be a good idea.  

 

Currently Diffmaker points to the idea you basically have the noise difference between the analog captures.  So we aren't furthering the discussion at this point in regards to what Mani and Mansr did in their get together.   

[wushuliu]

On 3/28/2018 at 1:48 PM, sandyk said:

It takes a lot of time to set up and do these things, as has been shown here at the various listening sessions I have been invited to attend and participate in. You also need to take at least one refreshment break to refresh your concentration.

 

A crucial observation for all those who live by the refrain 'but could you pass a blind test'. There is no Blind Test Store. You can't order one on Amazon. There is no Groupon. It takes Time, Effort, and Money. And if you have less of one, you've got to compensate with the other two.

 

I applaud what mansr and mani did, regardless of the outcome. Too bad I do not see some of the usual suspects with the usual comments after the results were posted.

[sandyk]

2 hours ago, esldude said:

I think we are cluttering up the thread talking about Diffmaker's inner workings.  I know we are talking about results with the files used.  Maybe a separate thread about the inner workings, pitfalls, and how much you can trust Diffmaker would be a good idea.  

 

 

 Dennis

That would be a very good idea.

I also tried using Diffmaker some years back, but I found it unreliable and too prone to crashing.

 

 For quite  a while it became the Peter-Mani thread too, where it should be the Mansr-Mani thread, with Mansr virtually being relegated to the sidelines , and to me at least, appearing to lose interest in the thread.

Perhaps some of Peter's content should have been sent initially via PM to Mani ? 

[fas42]

8 hours ago, psjug said:

If there is any problem in that band it seems to be the same on A and B.  Diffmaker is only showing file differences in the higher ultrasonics, where random noise does not subtract out.

 

I'm looking for patterns in what's going on ... and this sometimes needs several cycles at looking at what one has, over time - one jumps to an early wrong conclusion, or thought, which fizzles out - which doesn't mean discarding the data one has; rather, a revisiting is required, using another technique.

 

In the end it may turn out that the operation of the recorder itself is causing too much interference, masking or disturbing vital details - I have been aware of this on several occasions; just having the circuitry of the monitoring device active is altering the environment too much, and you lose what you're trying to measure. This always has to be considered, https://en.wikipedia.org/wiki/Observer_effect_(physics).

[semente]

On 18/04/2018 at 12:41 PM, manisandher said:

 

OK.

 

For my part, all can say is that all the XXHighEnd users around the world (without exception), using a whole range of DACs and gear, will readily believe that I heard differences in the sound of the music itself. Why? Because they hear the exact same differences.

 

Mani.

 

And in theory most of us should all be able to hear those differences by comparing the analogue captures, particularly now that you've indicated what to look for.

Is worth creating a couple of ABX files and run a poll?

[manisandher]

8 minutes ago, semente said:

And in theory most of us should all be able to hear those differences by comparing the analogue captures, particularly now that you've indicated what to look for.

 

This is what I heard in the A/B/X:

 

On 3/27/2018 at 10:01 PM, manisandher said:

I was listening to the piano transients - were they 'incisive' or 'blurred'/'soft'? With the female vocals, I was listening to her sibilance - acceptable, or too sharp and annoying.

 

Generally, it's much more about 'focus' than it is about the 'quantity' of any particular thing. For example, I didn't hear more bass, or more highs in one over the other, or anything like that.

 

I cannot discern any of this in the analogue captures, and am sure I'd fail an A/B/X miserably using them. Whatever caused the differences I heard (and can still hear clearly when listening directly) seems not to have been captured with the ADC we used.

 

Mani.

[manisandher]

10 hours ago, fas42 said:

In the end it may turn out that the operation of the recorder itself is causing too much interference, masking or disturbing vital details...

 

On 3/30/2018 at 8:19 AM, manisandher said:

The issue now is that no matter how accurately a diff program can find differences, it will likely not find consistent differences between the analogue outputs of the two playback means we used. That is, any differences it finds between 'A' and 'B' will likely be exactly the same sort of magnitude as the differences it finds within all the 'A's and within all the 'B's. (I suppose the differences will be > -96dB, owing to the 16-bit DAC used.) In which case, any such diff program, no matter how accurate, would totally fail the A/B/X test.

 

The only conclusion would be that it's measuring the wrong thing.

 

Mani.