Acourate and the unbalanced stereo setup

[jvvita]

Thanks. I will be eagerly waiting for your good news.

To be honest, what intrigued me most was the difference in frequency response when adding left and right post filter compared to the mono measurement

For now, just enjoying the music.

[Ecwl]

I just re-read what you wrote, jvvita.
You can't just add left and right channel in the frequency response.
You're supposed to use the add function in the time domain to add the Impulse response.
It should show what your measured mono response if you run it that way.

Think about it, this is a phase phenomenon. So if you just add the frequency responses up, of course you won't get a cancellation. You have to add them in the time domain to get the phase cancellation

[UliBru]

Mmh, the chart legend looks indeed a bit strange.
The test convolution creates TestL in curve1 and TestR in curve2.
Curve3 shows Sum12. It is a time domain addition of Pulses in curve1 and 2.

But now curve1 has a legend with content AdditionL+R. What has happened here?

And furthermore: is the mono recording done with or without correction? It looks like a raw measurement without correction.

[jvvita]

Yes. It was a time domain addition. Please ignore the legend. I see now that I made a mistake. I wanted to rename curve3, and ended up renaming curve 1.

The mono measurement is with the filters loaded, without any pre filtering.

[UliBru]

You should verify the test convolution first with a stereo logsweep measurement and active correction filters (stereo wav in LSR).
Then add the pulses in time domain.

[jvvita]

Yep. You are right.

Now TD Addition and Mono measurement (after the filters) match precisely.

Thank you once again. And eagerly waiting for the new features.

[Ecwl]

Jvvita, I just noticed you’re actually using method #1 Uli posted on July 13, 2021 to generate the 44Hz phase shift to cancel the 44Hz null.

But I find that you can get more optimal results by following Uli’s July 24, 2021 post (method #2). The underlying concept is that the reason why you have a left-right phase shift at 44Hz is because you probably have say a 30Hz phase shift on the left channel and a 60 Hz phase shift on the right channel. So if you measure the 30Hz phase shift on the left channel and then add it to the right channel in a prefilter and then add the right-channel 60Hz phase shift to the left channel, then both channels are going to be in phase and the 44Hz null will disappear. The reason why this approach is better than just adding a 44Hz phase shift to one channel is that sonically, the left and right channel sound more coherent in the bass region which you lose if you just add a 44Hz phase shift to one channel just to correct the frequency response.

[jvvita]

Thanks. I considered this other method, but I couldnt get around the spikes of the excess phase of my left channel (two spikes one at 56 and the other at 60hz). The allpass would have a Q of about 25.

To the right channel, I can generate an all pass of q=10 at 58hz that aligns perfectly with the excess phase "bump". But then again I dont know how to handle the two spikes (one near 20hz and the other near 100hz).

Any thoughts?

excess phase.png (19.46 KiB) Viewed 3288 times

[Ecwl]

Yeah. I don’t know. Would be good to see what Uli says.
I don’t think you need to do anything with the 95Hz phase shift since it’s not causing cancellation between left and right speakers.

My guess is that you need to add the 22Hz right channel phase shift to the left channel. But I really don’t know why both your left and right channel has all these phase shifts around 60Hz. I also don’t know why the phase shifts are so spikey. I wonder if that’s related to you using too high an EPW in Macro4. No idea.

Would be interesting to see Uli’s opinion on this.

[jvvita]

Yep. Challenging, to say the least. But, on the bright side, raw measurements return an early IACC of 81% and overall of 70%. Post correction, it shows 86% and 76% respectively. Sweet sounding