Blue or red pill?

[adamdea]

1 hour ago, jabbr said:

Where to start? So ... throwing out 1/2 the information ... ??‍♂️

 

Telling me that you cant see the differences that you can hear ??‍♂️

 

Lets see: we are looking at a phenom that is likely phase related but throwing out the phase info? ??‍♂️

 

This isn’t just theory, it’s math

No it's actually worse than that- it's telling you that you can't see the differences that you can't hear. Really.

[adamdea]

23 minutes ago, STC said:

 

Will use headphones with Foobar in pc and update later. 

Please use the Abx plugin.

[manisandher]

7 minutes ago, STC said:

 

File 3 and 4.

 

OK. Files 3 and 4 are 24/176.4 recordings of the DAC's analogue outputs going into the Tascam's analogue inputs ("analogue captures" for short). You may well hear very subtle differences between these two - I'm not sure I can.

 

Mani.

[manisandher]

33 minutes ago, adamdea said:

The thread is all over the place...

 

You're getting your knickers in a twist for no reason.

 

33 minutes ago, adamdea said:

... there are least two different points which are getting mixed up. 

1) is there an audible difference in the files. So far I see no evidence in favour. At least 2 negative ABx

2) if so does that correspond with any measurable characteristic.

 

@PeterSt and I have predicted, from pretty much the beginning of this thread, that:

1) there would be an audible difference in the A/B/X

2) there would be a) no measurable difference in the digital captures, and therefore b) no audible difference in the digital captures

3) there would be a) no meaningful measurable difference in the analogue captures (of the music track, not any test tones), and therefore b) no audible difference in the analogue captures

 

1) and 2) have been shown to be correct.

3a) looks to be correct, according to @testikoff and @esldude.

3b)... well, let's wait and see.

 

Mani.

[pkane2001]

4 hours ago, PeterSt said:

 

Huh ?

 

If an FFT shows differences but you can't perceive that, then you are deaf or your system is not on par.

Aparenty you think the other way around, somehow.

 

If an FFT does  not show differences, the sound can still be vastly different.

(I'd say that everybody knows this and it is a sad thing it is so)

 

The FFT shows some differences. I posted this yesterday. The differences rise close to an audible dB level, so I'm not sure why we keep saying there are no differences in the two captures.

 

Today, I used diffmaker to match up the A and B analog tracks and compute the difference. A similar set of frequencies appear in the delta (with differences), although with diffmaker the level is a bit higher than in my manual differencing process. Not sure why that is, I have to assume diffmaker is better at matching up using correlation technique.

 

Here's my manual spectrum of differences:

15 hours ago, pkane2001 said:

 

And here's one produced by diffmaker:

 

 

[manisandher]

2 minutes ago, pkane2001 said:

... I'm not sure why we keep saying there are no differences in the two captures.

 

Because:

 

17 hours ago, testikoff said:

Here you go:

 

- Delta closeup (frequency range 20...20000Hz; logarithmic scale):

 

And:

 

15 hours ago, esldude said:

... With sample rate drift correction (when it doesn't blow up) [Audio Diffmaker] shows a null depth of 94.5 db left and 92.3 db right.

...

I am not sure what could be left with nulls into the 90s with a 16 bit DAC with a noise in that same vicinity.

 

Mani.

[pkane2001]

2 minutes ago, manisandher said:

 

Because:

 

 

And:

 

 

Mani.

 

And this explains the -74dB signal at around 400-500Hz left over after subtracting A and B?

 

[adamdea]

7 minutes ago, manisandher said:

 

You're getting your knickers in a twist for no reason.

 

 

@PeterSt and I have predicted, from pretty much the beginning of this thread, that:

1) there would be an audible difference in the A/B/X

2) there would be a) no measurable difference in the digital captures, and therefore b) no audible difference

3) there would be a) no meaningful measurable difference in the analogue captures (of the music track, not any test tones), and therefore b) no audible difference

 

1) and 2) have been shown to be correct.

3a) looks to be correct, according to @testikoff and @esldude.

3b)... well, let's wait and see.

 

Mani.

My knickers are straighter than ever.

On 3), so far as we are discussing digital captures of the analogue output of the dac, I have not been able to discern a consistent position. If the captures yield an audible difference all well and good; if not then there is room for a number of theories as to why that may be the case.     

[manisandher]

8 minutes ago, pkane2001 said:

And this explains the -74dB signal at around 400-500Hz left over after subtracting A and B?

 

There isn't a -74dB signal at around 400-500Hz, according to Testikoff's analysis:

 

Mani.

[pkane2001]

5 minutes ago, manisandher said:

 

There isn't a -74dB signal at around 400-500Hz, according to Testikoff's analysis:

 

Mani.

 

I see. I found these frequencies doing the calculation two different ways, but who knows, maybe I did something wrong. I'll go back and recheck all the settings in diffmaker, although I got a similar result without it.

 

[manisandher]

1 minute ago, pkane2001 said:

I'll go back and recheck all the settings in diffmaker, although I got a similar result without it.

 

Testikoff was pretty thorough:

 

18 hours ago, testikoff said:

... just lined up samples the best I could on time axis (almost impossible to do with all the HF noise present), plotted their spectra & then calculated spectral delta.

 

Mani.

 

[pkane2001]

1 minute ago, manisandher said:

 

Testikoff was pretty thorough:

 

 

Mani.

 

That's exactly what I did manually. Then, I let diffmaker do the alignment. Both showed some left over frequencies in the delta. Diffmaker is a bit of a pain, some settings cause it to crash. I'm trying to find some settings that produce the best result without crashing. Not easy

 

[adamdea]

4 minutes ago, manisandher said:

 

Testikoff was pretty thorough:

 

 

Mani.

 

I can't follow his technique. AFAIK it is more usual to line up and substract in the time domain then take the spectrum of the time domain difference. I'm not clear how he has dealt with phase differences in the spectrum. It would be interesting to see what his time domain delta is.

[adamdea]

7 minutes ago, pkane2001 said:

That's exactly what I did manually. Then, I let diffmaker do the alignment. Both showed some left over frequencies in the delta. Diffmaker is a bit of a pain, some settings cause it to crash. I'm trying to find some settings that produce the best result without crashing. Not easy

 

I don't think it is the same as what you did. 

[mansr]

32 minutes ago, pkane2001 said:

Today, I used diffmaker to match up the A and B analog tracks and compute the difference. A similar set of frequencies appear in the delta (with differences), although with diffmaker the level is a bit higher than in my manual differencing process. Not sure why that is, I have to assume diffmaker is better at matching up using correlation technique.

 

Here's my manual spectrum of differences:

15 hours ago, pkane2001 said:

 

And here's one produced by diffmaker:

Remember this figure and those similar to it?

Those disturbances seem to coincide in frequency with peaks in the diffmaker plot.

[jabbr]

2 hours ago, PeterSt said:

So *now* you suddenly talk about speed = possible jitter measurement. But I never said that could not be done. That is only lacking equipment (OK, $).

OK, you didn't say jitter for real, but we could make it that. Anyway that is not an FFT which requires the bit depth and btw memory depth. AND speed.

The combination does not exist.

 

There are different measurements with different requirements and using different techniques.

 

For phase error (the type of jitter) there are at least 4 different techniques and can either be done in the analogue domain eg Herbert Rutgers’ device or the HP 3048a or digital domain eg “TimePod” or hybrid eh HP 5505. 

 

Need to look at the desired SNR, center frequency eg MHz vs GHz carrier as well as offset frequency resolution. 

 

The reason I acquired the parts for the HP 3048a — hardest part is a vintage CPU capable of running slow enough for the software (!!!) — is that it has a really terrific SNR as well as close-in phase error capability — now let’s say I have a project on my long stack — to create a digital version with an FPGA — and this digital version would be able to run alongside the DAC and measure DAC signals in real time — that would be the “holy grail” in terms of being able to see how, for example, your changes in software parameters affect jitter in different parts of the DAC...

 

okay but but this is a development project requiring resources and would someone buy that ($$$) ? How many would sales be? back to ROI — just not sure it would make sense as a stand-alone product ... but as a package in a $$$$$ system? ??‍♂️

 

anyways since this is hobby, my kids lacrosse/hockey/soccer games come first, as do concerts etc but I can post on my iPhone from the sidelines ?

[pkane2001]

5 minutes ago, adamdea said:

I don't think it is the same as what you did. 

 

Maybe I misunderstood what he did, then. Here's what I did (manually):

 

In Audacity, I tried to match the phase of the signals as close as possible. Zoomed in down to individual sample view, I made sure the peaks/valleys aligned on the same sample number in both captures. This was mostly already well aligned in the captures, but a bit of deleting of samples was needed to make the match better.

 

Then, I inverted one and added them together to produce the difference. Then, plotted the spectrum.

[adamdea]

8 minutes ago, pkane2001 said:

 

Maybe I misunderstood what he did, then. Here's what I did (manually):

 

In Audacity, I tried to match the phase of the signals as close as possible. Zoomed in down to individual sample view, I made sure the peaks/valleys aligned on the same sample number in both captures. This was mostly already well aligned in the captures, but a bit of deleting of samples was needed to make the match better.

 

Then, I inverted one and added them together to produce the difference. Then, plotted the spectrum.

As I have stated above- the normal technique AFAIK is to subtract in the time dmain and then take the spectrum. 

What Tesdtuikoff says he did is

 "lined up samples the best I could on time axis (almost impossible to do with all the HF noise present), plotted their spectra & then calculated spectral delta."

ie he took the [magnitude only?] fourier transform  of each and then subtracted to yeild a [magnitude only] fourier transform. By my reckoning under this methodology you are only looking at the total spectral content of each signal not comparing the signals at each point in time.

On that methodology two completely different files could come out the same. 

Using the more usual technique even a magnitude only FT will still respond to phase differences in the original signals 

@mansrcan you help here.....

[pkane2001]

1 minute ago, adamdea said:

ie he took the [magnitude only?] fourier transform  of each and then subtracted to yeild a [magnitude only] fourier transform.

 

Ah, that makes sense. I did something similar early on and found no visible differences in the spectra, which is why I went with matching up samples in the time domain and then subtracting the two signals and plotting the spectrum of the difference.

 

[jabbr]

3 hours ago, manisandher said:

 

It looks to me like @testikoff did the analysis adequately:

 

Wouldn't you agree?

 

If a higher resolution ADC, say 32/705.6, had been used, do you think the delta would have been any different in this plot?

 

Mani.

Mani,

 

I think @opus101 makes my point, looking at simple real only FFT of any resolution would miss the obvious case where a forward and time reversed sound file would obviously sound completely different but have the same real only (magnitude only) FFT. 

 

So no.

 

Also there are so many variations here that it’s hard to keep track of what it what

[STC]

1 hour ago, manisandher said:

 

OK. Files 3 and 4 are 24/176.4 recordings of the DAC's analogue outputs going into the Tascam's analogue inputs ("analogue captures" for short). You may well hear very subtle differences between these two - I'm not sure I can.

 

Mani.

 

foo_abx 2.0.4 report
foobar2000 v1.3.16
2018-04-17 22:46:08

File A: 3. analogue capture _ A.wav
SHA1: 9c7dbfe1ac5f00c6f3d3acd73c3d6a3fd1eaa59d
File B: 4. analogue capture _ B.wav
SHA1: 6c1ec457e27079d3e864c0396851e91185fcc982

Output:
DS : Primary Sound Driver
Crossfading: NO

22:46:08 : Test started.
22:46:43 : 01/01
22:48:58 : 02/02
22:49:24 : 03/03
22:49:48 : 04/04
22:50:02 : 04/05
22:53:11 : 04/06
22:53:40 : 05/07
22:53:49 : 05/08
22:54:01 : 05/09
22:54:21 : 06/10
22:54:21 : Test finished.

 ---------- 
Total: 6/10
Probability that you were guessing: 37.7%

 -- signature -- 
f1d262e5dee859f9f1671dc2492cf2b84b092c1a

[adamdea]

3 minutes ago, pkane2001 said:

 

Ah, that makes sense. I did something similar early on and found no visible differences in the spectra, which is why I went with matching up samples in the time domain and then subtracting the two signals and plotting the spectrum of the difference.

 

No surprises. No two analogue files null perfectly. Say they null to -90dB. Where does that take you until such time as some evidence emerges about there being an audible difference between the files? I have lost the thread as to whether there is any evidence as to whether the As and Bs are more like other As and Bs respectively than they are to each other.(either with test tones or music)

[adamdea]

13 minutes ago, STC said:

 

foo_abx 2.0.4 report
foobar2000 v1.3.16
2018-04-17 22:46:08

File A: 3. analogue capture _ A.wav
SHA1: 9c7dbfe1ac5f00c6f3d3acd73c3d6a3fd1eaa59d
File B: 4. analogue capture _ B.wav
SHA1: 6c1ec457e27079d3e864c0396851e91185fcc982

Output:
DS : Primary Sound Driver
Crossfading: NO

22:46:08 : Test started.
22:46:43 : 01/01
22:48:58 : 02/02
22:49:24 : 03/03
22:49:48 : 04/04
22:50:02 : 04/05
22:53:11 : 04/06
22:53:40 : 05/07
22:53:49 : 05/08
22:54:01 : 05/09
22:54:21 : 06/10
22:54:21 : Test finished.

 ---------- 
Total: 6/10
Probability that you were guessing: 37.7%

 -- signature -- 
f1d262e5dee859f9f1671dc2492cf2b84b092c1a

Thanks for having a go. I would regard that as a negative (but you seem to have started out well!). There's nothing to stop you having another go. I hope other people will have a try.

[testikoff]

The waveform delta of of my loosely aligned analogue capture 1 & 2 excerpts is not great, admittedly... The excerpts spectral delta, however, shows no substantial difference in spectrum within audible range, so regardless of small differences around 400Hz noted by others, they will be extremely hard to discern when the actual track excerpts are played back (not their refined null-test result produced by DiffMaker), IMHO...

[STC]

9 minutes ago, adamdea said:

Thanks for having a go. I would regard that as a negative (but you seem to have started out well!). There's nothing to stop you having another go. I hope other people will have a try.

 

During the trial, I got the first 4 wrong and all correct till 30 then 8 in a row were wrong. I suspect there is a slight different between the left and right phase. Otherwise, I would not perceive change in the image.