Beta-test: DeltaWave Null Comparison software

[pkane]

If I swap the reference and comp files sometimes the results are vastly different, is it an intended behavior?
View attachment 44245
View attachment 44246

Shouldn’t work this way, considering that all other results look similar. I’ve had it on my list to double check the linearity plot as it often gives results that are either too good or too poor. I’ll check into it.

 

[pkane]

If I swap the reference and comp files sometimes the results are vastly different, is it an intended behavior?
View attachment 44245
View attachment 44246

@bennetng :

Please test the new version 1.0.50. This should help with the linearity plot. I reduced the precision of the plot to make the visual appearance less chaotic. The sign of the non-linearity was causing a large discrepancy when swapping reference/comparison files, so now it displays the magnitude of the difference instead, which should be much more predictable and well-behaved.

For example:

 

[bennetng]

Improved a lot, thanks!

 

[Rigel]

I just found these two programs of yours and I'm surprised at your skills. Congratulations and thank you very much for your time. I have very limited knowledge and also the language, but I will try to investigate with them, especially with DeltaWave.

 

[exaudio]

I know there are quite a few Linux users on ASR so I thought I'd share that I was able to run @pkane's DeltaWave on Ubuntu using Wine. I didn't do much compatibility testing, but I was able to compare a couple of wav files without issue.

DeltaWave 1.0.50
Ubuntu 18.04.4 LTS
Wine 4.0.3 (64-bit prefix)

 

[pkane]

I know there are quite a few Linux users on ASR so I thought I'd share that I was able to run @pkane's DeltaWave on Ubuntu using Wine. I didn't do much compatibility testing, but I was able to compare a couple of wav files without issue.

DeltaWave 1.0.50
Ubuntu 18.04.4 LTS
Wine 4.0.3 (64-bit prefix)

View attachment 49361

That’s excellent! Can you please share the steps needed to get this running so others can try it?

Regards,

-Paul

 

[exaudio]

Can you please share the steps needed to get this running so others can try it?

Sure. I'm not an expert, but I can share what worked for me on the x64 version of 18.04.4 LTS.

First if you don't already have wine installed then install wine-stable 4.0.3 from Ubuntu's app store.
If Wine is installed and you have downloaded and extracted the DeltaWave installer, then you may be able to simply right-click on DeltaWaveSetup.exe and select 'Open With Windows Program Loader'.

If that doesn't work, then you might try a tool called Winetricks. Open a terminal and type the following:

cd "${HOME}/Downloads"
wget  https://raw.githubusercontent.com/Winetricks/winetricks/master/src/winetricks
chmod +x winetricks
bash ./winetricks

That last command should launch the Winetricks GUI. From there select 'Create new wineprefix' and create a new 64-bit prefix.

Next select 'Run explorer'

That should launch a familiar looking windows explorer window. Use that to browse to where ever you extracted DeltaWaveSetup.exe to. Double click it and it should launch the installer. Walk through the installer and you should be up and running.

One note--I did make an effort to use the newer Wine 5.0 available from https://www.winehq.org but, while DeltaWave would start up, it would freeze after clicking the 'Match' button. Your mileage may vary.

 

[Blumlein 88]

Sure. I'm not an expert, but I can share what worked for me on the x64 version of 18.04.4 LTS.

First if you don't already have wine installed then install wine-stable 4.0.3 from Ubuntu's app store.
If Wine is installed and you have downloaded and extracted the DeltaWave installer, then you may be able to simply right-click on DeltaWaveSetup.exe and select 'Open With Windows Program Loader'.

View attachment 49506

If that doesn't work, then you might try a tool called Winetricks. Open a terminal and type the following:

cd "${HOME}/Downloads"
wget  https://raw.githubusercontent.com/Winetricks/winetricks/master/src/winetricks
chmod +x winetricks
bash ./winetricks

That last command should launch the Winetricks GUI. From there select 'Create new wineprefix' and create a new 64-bit prefix.
View attachment 49510
Next select 'Run explorer'
View attachment 49511
That should launch a familiar looking windows explorer window. Use that to browse to where ever you extracted DeltaWaveSetup.exe to. Double click it and it should launch the installer. Walk through the installer and you should be up and running.

One note--I did make an effort to use the newer Wine 5.0 available from https://www.winehq.org but, while DeltaWave would start up, it would freeze after clicking the 'Match' button. Your mileage may vary.

Well I was very happy. Using the latter variation I managed to get Deltawave and Distort running on Manjaro linux (arch based). Except when I was ready to shut it down the only way to shut it down was to end the process in Ksysguard. Then the only way to get it to start again is to go back thru the installation process as if from scratch. So close. Darn it.

 

[pkane]

Well I was very happy. Using the latter variation I managed to get Deltawave and Distort running on Manjaro linux (arch based). Except when I was ready to shut it down the only way to shut it down was to end the process in Ksysguard. Then the only way to get it to start again is to go back thru the installation process as if from scratch. So close. Darn it.

At least on Ubuntu, all I needed to do to run it again was to find the executable. It should be under .wine/drive_c/Program Files (x86)/PKAudio/Distort (or DeltaWave). If I right click on the .exe there, I get the option to start it using Wine. No need to reinstall!

 

[GoldenOne]

Trying to use this to do a few tests but seemingly no matter what I do the software reports an enormous difference.
Do I have to align the tracks myself first?

Even copying the SAME wav file, then simply cutting the end 2 secs off file B, means deltawave then reports the two as basically completely different even though they are essentially the same file

 

[pkane]

Trying to use this to do a few tests but seemingly no matter what I do the software reports an enormous difference.
Do I have to align the tracks myself first?

Even copying the SAME wav file, then simply cutting the end 2 secs off file B, means deltawave then reports the two as basically completely different even though they are essentially the same file

What’s in the file, music or test signals and how long?

 

[GoldenOne]

What’s in the file, music or test signals and how long?

They're music.
Comparing audio diffmaker, which finds a 110dB correlated Null, and deltawave which only finds 85dB i'm not sure why they're so different.

I've uploaded the two files here: https://drive.google.com/drive/folders/1icXST5DS7XsaPF0gRwGCJy1mnARrC13N?usp=sharingThis is AFTER I trimmed both recorded files to the nearest sample as well. The difference still persists.

Also as mentioned, even if I took the same file, trimmed silence off the end of one, and then compared that with the original, deltawave reports massive differences despite the fact that they are actually the same file other than the length.

 

[pkane]

They're music.
Comparing audio diffmaker, which finds a 110dB correlated Null, and deltawave which only finds 85dB i'm not sure why they're so different.

View attachment 79039
View attachment 79040I've uploaded the two files here: https://drive.google.com/drive/folders/1icXST5DS7XsaPF0gRwGCJy1mnARrC13N?usp=sharingThis is AFTER I trimmed both recorded files to the nearest sample as well. The difference still persists.

Also as mentioned, even if I took the same file, trimmed silence off the end of one, and then compared that with the original, deltawave reports massive differences despite the fact that they are actually the same file other than the length.

Thanks for posting the files. Off-hand I can't tell what's going on, although I think it's related to the high sampling rate of the two files. In some way this is triggering some bug in the logic. If you go to settings and chose to resample the files to 48kHz, the result will be much closer to what you'd expect:

I'll take a look at what's causing it and will publish a fix.

 

[GoldenOne]

Thank you for the sample rate thing! Resampling to 96khz and also enabling a start/end LP filter at 20khz now seems to show much closer results to audio Diffmaker! Still seems to be slightly different (not sure why)

Not sure what could be causing the high sample rate issue, hopefully its not too tricky a fix! Thank you so much for this brilliant program!

 

[Blumlein 88]

Selecting Level EQ with 524K fft gets a correlated null of 102.5 db without resampling. What is left has that tilted tinny sound. That usually indicates a residual timing mismatch. Maybe high jitter?

 

[pkane]

Selecting Level EQ with 524K fft gets a correlated null of 102.5 db without resampling. What is left has that tilted tinny sound. That usually indicates a residual timing mismatch. Maybe high jitter?
View attachment 79088

I think @GoldenOne is on the right track. Just taking the original two files, with no resampling but with a 24kHz low-pass filter produces a much better result:

And by the way, if I enable drift compensation in Audio DiffMaker, here's the result (I was amazed it didn't crash for me, probably the first time that happened!)

>> Parameters: 2.678usec, 0.000dB rate adj=-0.0466 ppm.Corr Depth: 100.9 dB

Notice that all the parameters are in very close agreement with the DeltaWave result.

 

[GoldenOne]

Heya!
I just wanted to ask if you had the opportunity to look into the high sample rate issue?
I'm hoping to use deltawave to compare the analog outputs of a few different dacs for a video i'm making but would need to record and compare at a 384khz sample rate to minimise the effects of any digital filters in the chain.

Also, one suggestion for a future feature request, would be the ability to see an A-Weighted correlation value as well? To give a slightly better indication of how similar two files are taking into account human hearing/our perception.

Once again, thank you for this brilliant software!

 

[pkane]

Heya!
I just wanted to ask if you had the opportunity to look into the high sample rate issue?
I'm hoping to use deltawave to compare the analog outputs of a few different dacs for a video i'm making but would need to record and compare at a 384khz sample rate to minimise the effects of any digital filters in the chain.

Also, one suggestion for a future feature request, would be the ability to see an A-Weighted correlation value as well? To give a slightly better indication of how similar two files are taking into account human hearing/our perception.

Once again, thank you for this brilliant software!

Didn’t have time to look into it yet, I’ve been working on a different set of features related to periodic phase error detection and correction.

Isn’t the dBA RMS and correlated null the same thing as what you’re asking?

 

[GoldenOne]

Didn’t have time to look into it yet, I’ve been working on a different set of features related to periodic phase error detection and correction.

Isn’t the dBA RMS and correlated null the same thing as what you’re asking?

Ooh that sounds like a good addition! Looking forward to that.

Im getting a new adc (adi-2 Pro fs r) this week so I'll give high sample rate comparison another go once it arrives and post here. It could be that the problem is related to my interface(motu m2) ?

In any case I'm sure the reduction in jitter and improved overall performance of the adi-2 will help regardless, so hopefully high SR will work then.

The dba thing yep you're completely right, not sure how I completely missed that.......
Thank you!!

 

[GoldenOne]

Didn’t have time to look into it yet, I’ve been working on a different set of features related to periodic phase error detection and correction.

Isn’t the dBA RMS and correlated null the same thing as what you’re asking?

So I just got the ADI-2 Pro and gave deltawave another go.

Made two recordings @ 192khz on the ADC.

Depending on the settings, deltawave gave the following results:
192khz (Native): 61dB Null / -43dB diff
98khz (Downsampled): 115dB Null / -72dB diff
48khz (Downsampled): 143dB Null / -72dB diff

For comparison, audio diffmaker reports an 86dB correlation

I've uploaded deltawave reports, plus the original files here:
https://drive.google.com/drive/folders/1RY3QUqzibh8JhLpNF974tcFwSlDjzwFu?usp=sharing

Hopefully it can help track down what this bug is.
I'm not sure what the best way to get the most truly accurate result is. Do you have any advice?
Thanks!