DC filter effect on the null RMS metric

[Music1969]

Except in my case, it was computed automatically, in 15-20 seconds. When DW reports 1.5Hz or greater HP filter frequency, this means it didn't find a solution. As I said, I'll add the logic to give an error message for this in the future. For now, it seems something isn't set correctly in your configuration.

First, try to uncheck drift correct -- you don't need it and it takes a lot longer to do the initial match.

Then, try using all the same the settings I have here. Change FFT size for Spectrum to 1M, 32k doesn't have enough resolution below 1Hz to see or measure the details:
View attachment 442263

I uninstalled and reinstalled and copied your settings. I don't know which of the 2 fixed it but now working correctly

Suggestion for those settings to be the DW default ? Maybe not ticking the min phase box obviously but the rest should be default?

 

[pkane]

I uninstalled and reinstalled and copied your settings. I don't know which of the 2 fixed it but now working correctly

Suggestion for those settings to be the DW default ? Maybe not ticking the min phase box obviously but the rest should be default?

I suspect that Spectrum Size was the problem. Remember, this DC filter option is... optional and at this point, still experimental. Not everyone will want to have the need to resolve frequencies and phase below 1Hz.

 

[pkane]

I uninstalled and reinstalled and copied your settings. I don't know which of the 2 fixed it but now working correctly

Suggestion for those settings to be the DW default ? Maybe not ticking the min phase box obviously but the rest should be default?

I just updated the uploaded version 2.0.18 to automatically use 1M or larger fft size for determining DC filter offset. It will no longer use the Spectrum FFT size specified, unless it's 1M or greater. And it will set minimum phase option and filter size to 1M (if smaller) when you answer Yes to the prompt to update HP filter.

EDIT: also added an error message when the computation doesn't converge. Most likely issue in this case is that there was no minimum phase DC filter in the loopback path, or its cutoff was well above 2Hz.

 

[Music1969]

@pkane I've hit process HP filter freq numerous times and get a different number each time

Is it the clock drift enabled doing it? I'll disable (per your advice)

Havent updated to new version

 

[pkane]

@pkane I've hit process HP filter freq numerous times and get a different number each time

Is it the clock drift enabled doing it? I'll disable (per your advice)

Havent updated to new version

After you add an HP minimum phase filter the first time, you've modified the the match computation since reference is now minimum phase. To get the same result, clear the HP filter value, turn off filter, or set the filter value to 0. Then ask for DC filter offset again. Otherwise, you're computing DC filter after applying another DC filter, ... after another, etc.

 

[Music1969]

After you add an HP minimum phase filter the first time, you've modified the the match computation since reference is now minimum phase. To get the same result, clear the HP filter value, turn off filter, or set the filter value to 0. Then ask for DC filter offset again. Otherwise, you're computing DC filter after applying another DC filter, ... after another, etc.

View attachment 442274

Ahhh got it. Yes makes sense. Retrying now

 

[Music1969]

Here's what I get with the same files. And notice the shape of the uncorrected phase -- that's an indication of minimum phase HP filter used in the recording.

Same settings as you and same SSL2 Mk2 .wav but different value to you

 

[pkane]

Same settings as you and same SSL2 Mk2 .wav but different value to you

View attachment 442276

View attachment 442277

It's very nearly the same, the difference is 0.05Hz -- that's about the precision of 1M FFT (bin size of 0.04Hz).

I was computing mine based on the right channel, so likely slightly different data and noise pushed the result over by one bin. In either case, the result is likely the same null value. Try both values and see.

 

[Music1969]

It's very nearly the same, the difference is 0.05Hz -- that's about the precision of 1M FFT (bin size of 0.04Hz).

I was computing mine based on the right channel, so likely slightly different data and noise pushed the result over by one bin. In either case, the result is likely the same null value. Try both values and see.

all good now, thanks for the help. time to re-run a lot of interfaces

 

[pkane]

all good now, thanks for the help. time to re-run a lot of interfaces

Thanks for testing! I'm going to make a few changes based on your feedback and publish a new version. This will clear the HP filter before doing any new DC filter offset computations.

 

[Music1969]

@pkane looking at delta phase plots before and after mp hp filter, here are same for SSL2+ Mk2.

Main phase improvement is under 100Hz.

How would you interpret overall results?

Without mp hp filter


With mp hp filter

 

[pkane]

Thanks for testing! I'm going to make a few changes based on your feedback and publish a new version. This will clear the HP filter before doing any new DC filter offset computations.

That's now updated.

 

[pkane]

@pkane looking at delta phase plots before and after mp hp filter, here are same for SSL2+ Mk2.

Main phase improvement is under 100Hz.

How would you interpret overall results?

Without mp hp filter
View attachment 442280

With mp hp filter

View attachment 442282

That's what I expect a DC filter will do, and DeltaWave will correct. You can see that the phase is nearly linear up to 10kHz after the correction. And null RMS has improved from -40 to -48dB. Not an insignificant change, and it does indicate that a minimum phase DC filter was in the path.

 

[Music1969]

One thing I dont get with the gearspace method, looking at spectrum of delta below.

its mostly better than -60dB difference across the spectrum.

The -39dB RMS seems pretty harsh

I guess thats where PK Metric is more realistic

 

[pkane]

One thing I dont get with the gearspace method, looking at spectrum of delta below.

its mostly better than -60dB difference across the spectrum.

The -39dB RMS seems pretty harsh

View attachment 442283

Remember that spectrum is an average over the whole 2 minute recording. There are likely peaks and valleys that far exceed the numbers you see on the average plot.

 

[Music1969]

DeltaWave v2.0.18: https://app.box.com/s/hlglgy3aj39rjfxu343fho4t0cdrq9v8

same link has latest updates?

Remember that spectrum is an average over the whole 2 minute recording. There are likely peaks and valleys that far exceed the numbers you see on the average plot.

fair point

 

[pkane]

same link has latest updates?


fair point

Yes, re-download from the same link. Uninstall the previous version, first, as I kept it at v2.018. Then install the new one.

By the way, try to correct DC blocking filter in this one, for fun

Antelope Pure2
Touf on Pragmatic Audio

 

[Music1969]

Yes, re-download from the same link. Uninstall the previous version, first, as I kept it at v2.018. Then install the new one.

By the way, try to correct DC blocking filter in this one, for fun

Antelope Pure2
Touf on Pragmatic Audio

Holy moly. Well it starts off really good but gets a huge improvement

Plot just happens to be on DF metric but can see difference RMS at the bottom of each

Without mp hp filter


With

 

[pjug]

To simplify and speed up the process of finding the correct DC minimum filter setting, I added a menu option to automatically compute the best approximation to filter frequency.

This is really great. Thank you!

How do you determine best frequency, or what metric are you optimizing for?

edit: I duplicated the results using your RME loopback file. Computed the 0.0566Hz filter and got -84.85dB null. But then looking at the delta phase it looked like it could possibly be flattened even more so I used a 0.04Hz filter and got a -85.42dB null.

 

[pkane]

This is really great. Thank you!

How do you determine best frequency, or what metric are you optimizing for?

edit: I duplicated the results using your RME loopback file. Computed the 0.0566Hz filter and got -84.85dB null. But then looking at the delta phase it looked like it could possibly be flattened even more so I used a 0.04Hz filter and got a -85.42dB null.

The solver looks for a filter frequency that produces the most linear phase response.

The resolution of the solver is 1 FFT bin, which for 1M filter is 0.04Hz. To get a higher precision, the solver would need to use a larger filter. I'd also have to decrease the tolerance to below 0.04Hz. That's doable, but it will slow down the performance of the solver by quite a bit.