DeltaWave null-testing audio comparator (beta)

[TomCapraro]

The software is great, and it works very well.
It would be wonderful if you could implement an overlay function so that you could save the result of the spectrum, then be able to change the parameters and superimpose it.
A sincere thanks to Paul for what he is doing.😊

[TomCapraro]

urgent ... the "non linear EQ" function does not work.
It does not align the frequency response.😕😌

[TomCapraro]

Try to compare the left channel with the right channel.
Then apply the EQ function, it is not aligned with me.
I add the file.

Senza titolo.WAV

[TomCapraro]

With my file working now, I'm doing some more tests

[TomCapraro]

All right, at least following a lineup that I keep preserved to test the software.
If I encounter any problems I will report it.
Thanks for the speed in solving the problem.😊

[TomCapraro]

Only now have I noticed surprise.
The overlay plot ... !!!

[TomCapraro]

💪✌️

[TomCapraro]

8 hours ago, Arpiben said:

Hi Paul,

 

Do you mind having a look at phase delta artefacts/noises ? It is a minor issue but Delta Phase result is not steady at high frequencies.

This is not specific to actual release and was present before.

Comparing two identical files and pressing successively SHOW leads to slight different results at every sequence/press.

For example:

 

Load this file and make the L-R comparison with it. I like the delta of the phase.

Senza titolo.wav

[TomCapraro]

Paul, because for a simple misalignment of 7 samples with 10khz frequency does not make the correct difference?
Below the 7 phase-shifted file between left and right channel.

10khz 7 samples.wav

[TomCapraro]

Another little problem, with the file you will find a problem on the delta of spectra. If you compare the left channel with the right it provides bit perfect but with a delta of spectra wrong with respect to the comparison between left and left.

problem delta of spectra.wav

[TomCapraro]

1 hour ago, pkane2001 said:

 

Hi Tom,

 

A simple sine-wave can't be properly aligned using DeltaWave because every cycle looks exactly like any other cycle anywhere in the waveform. The method of determining clock drift relies on the fact that a correct delay/offset can be measured at multiple parts of the waveform. Since there are as many ways to align the waveform as there are cycles when using a sine-wave, DW can't compute clock drift properly. I did promise a long time ago to look into handling simple waveforms like this, but it's not going to work right with the current version.

 

You can still try to align this by trying just the non-linear phase EQ, but turn off all other correction options, especially drift correction. For this, the file sample size should be much longer compared to the FFT size specified in the non-linear EQ settings. Or, you can use the Manual Adjustment window and enter the correct offset directly.

Okay, good to know.

[TomCapraro]

1 hour ago, pkane2001 said:

 

I'm not sure I understand the problem. Here's the delta of spectra between L and L channels of this file:

 

And here's Left and Right channel comparison:

 

Are you seeing something different than this?

 

Try MATCH rather than SHOW
The two signals are the same, and should provide bit perfect.
Only the right is out of 7 samples.

[TomCapraro]

7 minutes ago, pkane2001 said:

 

Ah, I understand. That's the same problem as we discussed earlier: the alignment will not work with a simple sine wave because DW can't find any unique patterns in the two waveforms.

 

Yes Paul, but here the problem arises with two Pink Noise signals.

[TomCapraro]

If instead I do the alignment through Audacity and / or Adobe Audition and extract the difference file, uploading it to DW gives me this delta spectra

[TomCapraro]

5 minutes ago, pkane2001 said:

 

I'm still not seeing the same behavior here. But I'm working on an updated version of DW, so maybe it's related to the fix I'm putting in for the issue Arpiben reported where phase and spectrum delta could show some small errors, despite being bit perfect. I'll post it here soon, maybe you can re-test it then.

 

 

 

Ok, 😉

[TomCapraro]

Great but ... I don't see any link to download the update ... !!!😁🤣

[TomCapraro]

I had a problem with the password to log in and I couldn't thank Paul right away for the good work he did.
Now the problems seem to be solved.

[TomCapraro]

It seems to me an excellent solution.
I will do some tests by tomorrow.
Paul, for complex signals do you advise to deactivate the function since it is activated by default?

[TomCapraro]

Offsets that are too long are not aligned with Paul?
is there a pre-defined sample window?

[TomCapraro]

2 hours ago, pkane2001 said:

 

Deactivate this for complex signals.  Only turn this on when you know that you are measuring a simple periodic waveform.

 

When enabled, this will not do any clock drift corrections and will do a much simpler cross-correlation and offset computation to find the fractional offset. The reason is that phase slope cannot be accurately computed from a single tone or even a few tones.

Paul, you say you don't use simple alignment with complex signals ... instead I get almost theoretical null results with the use of simple alignment on a pink noise loopback then comparing the digital file with the one acquired in analog.
I believe that we must always try as a valid alternative.

[TomCapraro]

22 hours ago, pkane2001 said:

 

The simplest way to eliminate the influence of an ADC on frequency response is to record the source twice, inverting ADC phase between the captures. Then add the two captures and divide by two. This does have the unfortunate effect of reducing noise by √2, but harmonic distortions should stay constant.

Hi Paul, can you explain better what you mean by this process? maybe it's something I do as well but ... when you talk about inverting the ADC add and then split I can't translate it

[TomCapraro]

While you were having coffee and thought how to get a great null-test by looping a dac and adc, I started drinking coffee and thinking about myself.😁
Since Paul introduced an alignment function for periodic signals, I created a pseudo-periodic 64k white noise signal.
Activating the alignment function, with this signal, I get an amazing null ... and we see it ...

 

Observe the values ...                                                                       

[TomCapraro]

The settings used ...

[TomCapraro]

And here when I disable the alignment of periodic signals ...

 

 

[TomCapraro]

I couldn't get more than that.
How does it look like?
Do you want me to place the signal so you can try it too?