Beta-test: DeltaWave Null Comparison software

[pkane]

Hi pkane, thanks for developing this, it's looks so incredibly useful and promising to shed light on perceived difference in digital audio reproduction..

I also posted on your website but I seem to perhaps have run into a bug?

I seem to have a problem with measuring the R channel exclusively, I always get completely off the chart delta phase results. I did all kinds of cross checking with different dacs, output modes, feeding the R out into the L of my ADC, swapping the channel in my test files and using different reference test files, but everything seems to point to a software bug. Also in stereo mode, I don't get the problem either. Could you please let me know if you can reproduce or if you need debug logging? It never happens on channel L.

Audio data on channel R

View attachment 449169

Same exact data on same ADC input from channel L (multiple DACs)

View attachment 449170

Thanks again

Probably not a bug. You need to select both, the reference and the comparison file channels to run the comparison. Both should be set to R if you want to compare right channels. If Comparison is set to R and Reference is set to L, you're comparing left channel to right, which will result in a large phase error.

Edit: Oh, and if you're using the built-in recorder in DeltaWave, you'll need to select 'stereo' for channels in the recorder, otherwise only one channel mono will be recorded.

 

[kieko]

I can't get it to work.. if I select R/R in the recorder it switches ref/compare to RR but this is the result:

no issues with the L channel

 

[pkane]

I can't get it to work.. if I select R/R in the recorder it switches ref/compare to RR but this is the result:
View attachment 449194

no issues with the L channel

Did you try recording stereo instead of specific channel? You can then use left or right selection in the main window

 

[kieko]

I did and seems to work but I'm having instances where one of the channels has a much different delta (not dac dependent) and the phase is off again. I'll see if I can reproduce tomorrow but at least for me the right channel has this problem 100% of the times with the recorder.

Also I get these weird patterns at the end in the spectrogram (they are like weird spikes in the delta waveform, likely a playback issue with the recorder?)

 

[pkane]

I did and seems to work but I'm having instances where one of the channels has a much different delta (not dac dependent) and the phase is off again. I'll see if I can reproduce tomorrow but at least for me the right channel has this problem 100% of the times with the recorder.

Also I get these weird patterns at the end in the spectrogram (they are like weird spikes in the delta waveform, likely a playback issue with the recorder?)

View attachment 449220

Possibly playback or a recording issue. Possibly an issue with the PC CPU not being able to handle the load, but it's strange that it happens only in one channel. You can examine the waveforms to see what this error looks like which may give you an idea of the source.

 

[kieko]

doesn't seem related to the recorder, when I compare any 2 tracks I get these 3 spikes at the end in the delta that must me skewing the results?

 

[pkane]

doesn't seem related to the recorder, when I compare any 2 tracks I get these 3 spikes at the end in the delta that must me skewing the results?


View attachment 449253

so these are in both channels and with any track ? I’ve not seen anything like this before.

 

[pkane]

doesn't seem related to the recorder, when I compare any 2 tracks I get these 3 spikes at the end in the delta that must me skewing the results?


View attachment 449253

If you know how many seconds at the end include the three spikes, you can just tell DeltaWave to trim that many seconds off the end.

 

[kieko]

I did some more testing and my guess is that it's something that derives from the drift computation, if I compare 2 identical files for example I don't get it, but comparing a wav with it's mqa counterpart I get the 3 bands:

so this has nothing to do with the recordings or recorder

 

[pkane]

I did some more testing and my guess is that it's something that derives from the drift computation, if I compare 2 identical files for example I don't get it, but comparing a wav with it's mqa counterpart I get the 3 bands:

View attachment 449362

so this has nothing to do with the recordings or recorder

How long is your recording and at what sample rate? What is the test signal? Drift calculation needs at least a few minutes of recording and music as the test signal.

 

[kieko]

about 1 minute, but isn't the drift calculation absolutely fine? It's just somehow adding some audio spikes at the end every single time, I think the data is only partially skewed by these, the results for the most part make sense. I think it's easy to reproduce?

44.1/16

 

[pkane]

about 1 minute, but isn't the drift calculation absolutely fine? It's just somehow adding some audio spikes at the end every single time, I think the data is only partially skewed by these, the results for the most part make sense. I think it's easy to reproduce?

44.1/16

Cant tell what was happening, but the drift calculation can be skewed by noise, which is why a longer recording works better — noise is averaged out. If you want, share the reference and one of the recordings and I can take a look.

 

[kieko]

Sent you the test clips / reports in pvt

 

[danadam]

The description of "Δ Spectrogram" tab when hovering the cursor says:

Spectrogram of the Delta of two waveforms

and the description in the documentation says:

FFT power spectrum of the difference of the two waveforms (the difference computed in time domain, then Spectrogram is computed)

Is that correct? Because given the scale (+/- around zero) it feels more like "Δ of Spectrograms".

Btw. "Δ of Spectra" has no description when hovering the cursor and there is no description in the documentation.

 

[pkane]

The description of "Δ Spectrogram" tab when hovering the cursor says:

and the description in the documentation says:

Is that correct? Because given the scale (+/- around zero) it feels more like "Δ of Spectrograms".

Btw. "Δ of Spectra" has no description when hovering the cursor and there is no description in the documentation.

Delta of spectra is the difference of the two spectrum plots. I may need to adjust the description on the delta of spectrograms, as that is literally the same type of a calculation, except shown with the added time dimension.

 

[pkane]

A new test version of DeltaWave is available (v2.0.21): https://app.box.com/s/hlglgy3aj39rjfxu343fho4t0cdrq9v8

This version includes a change to address some of the reported issues with the auto-trim function and an improved clock drift calculation that should work a bit faster and produce more precise results in the presence of a large drift.

 

[Rantapossu]

A new test version of DeltaWave is available (v2.0.21): https://app.box.com/s/hlglgy3aj39rjfxu343fho4t0cdrq9v8

This version includes a change to address some of the reported issues with the auto-trim function and an improved clock drift calculation that should work a bit faster and produce more precise results in the presence of a large drift.

Thanks!

The phase plot seems still to be updated prior to DC filter removal.

The phase plot right after the matching:

After pressing the Refresh

-button:

 

[pkane]

Thanks!

The phase plot seems still to be updated prior to DC filter removal.

View attachment 456488

The phase plot right after the matching:

View attachment 456489


After pressing the Refresh View attachment 456491 -button:

View attachment 456493

Yes, thanks for reminding me! I thought I had made that change. It would appear the change didn't work

 

[pkane]

Thanks!

The phase plot seems still to be updated prior to DC filter removal.

View attachment 456488

The phase plot right after the matching:

View attachment 456489


After pressing the Refresh View attachment 456491 -button:

View attachment 456493

Hi @Rantapossu,

Can you please try this new version? It attempts to update all the plots after a DC filter correction is applied, so should reflect the new phase delta: https://app.box.com/s/hlglgy3aj39rjfxu343fho4t0cdrq9v8

Uninstall previous version 2.0.21 as this has the same version number.

 

[Rantapossu]

Hi @Rantapossu,

Can you please try this new version? It attempts to update all the plots after a DC filter correction is applied, so should reflect the new phase delta: https://app.box.com/s/hlglgy3aj39rjfxu343fho4t0cdrq9v8

Uninstall previous version 2.0.21 as this has the same version number.

Thanks again!

Works ok now!