Beta-test: DeltaWave Null Comparison software

[pkane]

I mean, the same result happens no matter whether it's loopback or even just applying Pro-Q3 filter or anything else useful.

So for example, take the GS original WAV, apply a single Pro-Q3 instance (probably the single most used/respected EQ plugin in the Music production world right now) with a 5dB (24dB/octave) cut at 15khz and compare:

L - diff -59.02dB [-63.42dBA], PKMetric RMS=-56.8dBr
Stereo - diff -34.24dB [-34.19dBA], PKMetric RMS=-43.3dBr

OK, 15kHz is still within audible range, so it will have some effect on audibility (PKMetric) and certainly on the RMS null metric since that includes the full bandwidth.

You can specify a Low-pass filter in DeltaWave set to 15kHz to see how similar/different the two are below that frequency, but I also suspect that the filters in Pro-Q3 are IIR and not particularly precise, so might affect lower frequencies, below 15kHz, as well as, phase. Post a spectra delta and phase delta plots to see what effect these filters have on the waveform. I wouldn't be surprised if the Pro-Q3 also resamples the audio as part of the internal processing, a lot of "pro" filters do this automatically.

 

[matpowel]

Did you test the R channel too?

Yes, essentially identical to the L results (within 1dB of all numbers).

R - diff -59.84dB [-63.42dBA], PKMetric RMS=-55.7dBr

Have you tried it yourself? Are you saying the same process doesn't give the same results? What would you all propose could be the cause? Are you suggesting either Logic Pro or Pro-Q3 are causing stereo alignment issues? What is the claim here?

I admire the confidence here, but I find it waaaaayyyy more believable that the DW algorithm for stereo diffing is flawed rather than Apple, FabFilter and all the converters I've tested etc ALL doing something weird to the audio, honestly speaking :-D

For now I'll just be using single channel results.

 

[matpowel]

OK, 15kHz is still within audible range, so it will have some effect on audibility (PKMetric) and certainly on the RMS null metric since that includes the full bandwidth.

You can specify a Low-pass filter in DeltaWave set to 15kHz to see how similar/different the two are below that frequency, but I also suspect that the filters in Pro-Q3 are IIR and not particularly precise, so might affect lower frequencies, below 15kHz, as well as, phase. Post a spectra delta and phase delta plots to see what effect these filters have on the waveform. I wouldn't be surprised if the Pro-Q3 also resamples the audio as part of the internal processing, a lot of "pro" filters do this automatically.

The issue we are addressing here is the large difference between the "L", "R" and "Stereo" results. This seems unexpected to me and the claim was made by @Grooved that there shouldn't be a large difference and you said that I must have something wrong with my setup. I'm just trying to understand if the "stereo" setting is useful and if so what for, given the massive difference between its result and the individual results of the "L" and "R" channel.

So far I haven't seen that anyone else has done any actually useful testing of it in the same way that I have to validate whether you can see the same thing. It would be really useful to know if anyone can replicate the same behavior (using any plugin or stock EQ) or really any kind of processing.

 

[pkane]

The issue we are addressing here is the large difference between the "L", "R" and "Stereo" results. This seems unexpected to me and the claim was made by @Grooved that there shouldn't be a large difference and you said that I must have something wrong with my setup. I'm just trying to understand if the "stereo" setting is useful and if so what for, given the massive difference between its result and the individual results of the "L" and "R" channel.

So far I haven't seen that anyone else has done any actually useful testing of it in the same way that I have to validate whether you can see the same thing. It would be really useful to know if anyone can replicate the same behavior (using any plugin or stock EQ) or really any kind of processing.

Stereo is the same as doing all the computations on the left channel, then applying the same corrections to the right channel. There are no other "Stereo" results, just the L channel comparison results. What I see is a huge clock drift in your results which tells me that something isn't right in your comparison or settings. Maybe share the loopback stereo recording file(s), and I can take a look.

 

[Grooved]

Yes, essentially identical to the L results (within 1dB of all numbers).

R - diff -59.84dB [-63.42dBA], PKMetric RMS=-55.7dBr

Have you tried it yourself? Are you saying the same process doesn't give the same results? What would you all propose could be the cause? Are you suggesting either Logic Pro or Pro-Q3 are causing stereo alignment issues? What is the claim here?

I admire the confidence here, but I find it waaaaayyyy more believable that the DW algorithm for stereo diffing is flawed rather than Apple, FabFilter and all the converters I've tested etc ALL doing something weird to the audio, honestly speaking :-D

For now I'll just be using single channel results.

Thanks for the answer. And no, I've used a lot DeltaWave but didn't tried your test yet, I will when I get a moment as I have this file and also Pro-Q3

The issue we are addressing here is the large difference between the "L", "R" and "Stereo" results. This seems unexpected to me and the claim was made by @Grooved that there shouldn't be a large difference and you said that I must have something wrong with my setup. I'm just trying to understand if the "stereo" setting is useful and if so what for, given the massive difference between its result and the individual results of the "L" and "R" channel.

So far I haven't seen that anyone else has done any actually useful testing of it in the same way that I have to validate whether you can see the same thing. It would be really useful to know if anyone can replicate the same behavior (using any plugin or stock EQ) or really any kind of processing.

I said there shouldn't be such a difference because I missed the part where you said these results were after applying a EQ cut, so it was not about the same thing (I thought it was the results of a D-A-D loopback compared to original.

 

[matpowel]

I said there shouldn't be such a difference because I missed the part where you said these results were after applying a EQ cut, so it was not about the same thing (I thought it was the results of a D-A-D loopback compared to original.

I've seen the exact same problem for everything I've tested, including D-A-D loopbacks.

 

[matpowel]

Stereo is the same as doing all the computations on the left channel, then applying the same corrections to the right channel. There are no other "Stereo" results, just the L channel comparison results. What I see is a huge clock drift in your results which tells me that something isn't right in your comparison or settings. Maybe share the loopback stereo recording file(s), and I can take a look.

Original GS file: https://www.dropbox.com/scl/fi/u5ke...nal2.wav?rlkey=wjsjrpeat269b5bxxxpb06h55&dl=0

Processed file (Pro-Q3 only, no conversion, only DAW internal plugin processing so there shouldn't be any "clocks" involved): https://www.dropbox.com/scl/fi/pxo4...-17k.aif?rlkey=vkdeppi41odjy6kymorhdwd74&dl=0

 

[pkane]

Original GS file: https://www.dropbox.com/scl/fi/u5ke...nal2.wav?rlkey=wjsjrpeat269b5bxxxpb06h55&dl=0

Processed file (Pro-Q3 only, no conversion, only DAW internal plugin processing so there shouldn't be any "clocks" involved): https://www.dropbox.com/scl/fi/pxo4...-17k.aif?rlkey=vkdeppi41odjy6kymorhdwd74&dl=0

Thanks, Matt. I am able to reproduce your result with the captured file... but only in AIF format. Reading both channels from AIF appears to be broken, while reading one or the other seems to be fine.

Here I took your loopback file, converted it to WAV in Audacity, and then measured it as Stereo in DW. Results are the same as for L or R channels, separately, as one would expect:

I'll take a look later to see why the two channel AIF reader is misbehaving.

 

[matpowel]

Thanks, Matt. I am able to reproduce your result with the captured file... but only in AIF format. Reading both channels from AIF appears to be broken, while reading one or the other seems to be fine.

Oh fascinating, so we found the bug but it's in your AIFF stereo parsing. Well at least that makes sense why you and @Grooved weren't seeing the same issue. Will you get a chance to fix the bug? I think in general stereo comparison for a stereo loopback is super useful as it also compares any weird channel specific or stereo phase issues in conversion or whatever is being tested.

Thanks for pursuing it.

 

[Blumlein 88]

Oh fascinating, so we found the bug but it's in your AIFF stereo parsing. Well at least that makes sense why you and @Grooved weren't seeing the same issue. Will you get a chance to fix the bug? I think in general stereo comparison for a stereo loopback is super useful as it also compares any weird channel specific or stereo phase issues in conversion or whatever is being tested.

Thanks for pursuing it.

Paul is usually pretty quick with the fixes, and I don't remember a single bug he hasn't fixed yet. So give him a bit of time.

 

[matpowel]

Paul is usually pretty quick with the fixes, and I don't remember a single bug he hasn't fixed yet. So give him a bit of time.

No rush/pressure here, I'm fine either converting to WAV or only comparing L or R separately for now.

 

[Grooved]

Oh fascinating, so we found the bug but it's in your AIFF stereo parsing. Well at least that makes sense why you and @Grooved weren't seeing the same issue. Will you get a chance to fix the bug? I think in general stereo comparison for a stereo loopback is super useful as it also compares any weird channel specific or stereo phase issues in conversion or whatever is being tested.

Thanks for pursuing it.

I just finished to test, but I found the bug was found, great.

And indeed, I had no bug at all using WAV. I did your test (except I'm not sure if we had the exact same setting in PRO-Q 3 (mine was on Zero Latency mode for example), and here are the results:

L - diff -48.5dB [-51.42dBA], PKMetric RMS=-48.3dBr
R - diff -49.01dB [-51.58dBA], PKMetric RMS=-48.3dBr
Stereo - diff -48.5dB [-51.42dBA], PKMetric RMS=-48.3dBr (like I said above, it gives the same result than the worst result between L and R)

 

[pkane]

Oh fascinating, so we found the bug but it's in your AIFF stereo parsing. Well at least that makes sense why you and @Grooved weren't seeing the same issue. Will you get a chance to fix the bug? I think in general stereo comparison for a stereo loopback is super useful as it also compares any weird channel specific or stereo phase issues in conversion or whatever is being tested.

Thanks for pursuing it.

Yep, I tested all the basic functionality with AIF, but not Stereo mode, and it seems, nobody else tried Stereo with it until now. I'll let you know when a fix is available.

 

[pkane]

DeltaWave version 2.0.10 is now available for download. This is to address the issue with AIF/AIFF files when processed in stereo mode (@matpowel - thanks for reporting!)

Changes in 2.0.10​

• Fix: issue when matching AIF/AIFF format files in Stereo mode
• Change: removed 32-bit version download option from the website

 

[Grooved]

DeltaWave version 2.0.10 is now available for download. This is to address the issue with AIF/AIFF files when processed in stereo mode (@matpowel - thanks for reporting!)

Changes in 2.0.10​

• Fix: issue when matching AIF/AIFF format files in Stereo mode
• Change: removed 32-bit version download option from the website

Thanks again @pkane !
I forgot to ask you if it would be possible to update the ASIO library like you did for MTA?
When one use AD or DA converters with only ADAT In or Out, it's needed to link them to an audio interface, and it's often not possible to select the ADAT channels as L or R

 

[pkane]

Thanks again @pkane !
I forgot to ask you if it would be possible to update the ASIO library like you did for MTA?
When one use AD or DA converters with only ADAT In or Out, it's needed to link them to an audio interface, and it's often not possible to select the ADAT channels as L or R

You mean the support for a larger number of input/output channels? Sure, I can do that.

 

[Grooved]

You mean the support for a larger number of input/output channels? Sure, I can do that.

Yes, exactly like it was added to MTA

 

[matpowel]

DeltaWave version 2.0.10 is now available for download. This is to address the issue with AIF/AIFF files when processed in stereo mode (@matpowel - thanks for reporting!)

Changes in 2.0.10​

• Fix: issue when matching AIF/AIFF format files in Stereo mode
• Change: removed 32-bit version download option from the website

Wow you don't mess about @pkane, thanks I'll download tomorrow.

 

[PO3c]

Any special trick needed to have the Phase delta plot function on Win10? Both 2.0.8 and 2.0.10 tested with same problem.

2023-11-08 01:27:21.2802|INFO|Wave.WaveForm|---- Phase difference (full bandwidth): 42,0426532979263°
2023-11-08 01:27:21.2802|INFO|Wave.WaveForm| 0-10kHz: 13,93°
2023-11-08 01:27:21.2802|INFO|Wave.WaveForm| 0-20kHz: 25,53°
2023-11-08 01:27:21.2802|INFO|Wave.WaveForm| 0-24kHz: 42,04°

 

[Grooved]

Any special trick needed to have the Phase delta plot function on Win10? Both 2.0.8 and 2.0.10 tested with same problem.

Hi,
never got this problem with W10, I just checked to be sure and the main difference I see is that I never got a Phase Offset result like that (-730,380405ms), it's more +/- 0.00xxxx
ms