Auto eq for rephase and left and right help

[Andrew Slater]

Hi guys I have a sort of complicated question in regards to using rew and rephase together and making correction filters for left and right with unequal path lengths and in a tiny room with wheels and an engine.

Ive been using rew for a couple years now, getting to know a lot of the controls, wouldn’t say I’m a expert at any of them

My question is how to I get the left and right impulses to match each other so there’s no de-correlation from having to have different eq settings on left and right?

In rephase and in rew I can only do one channel at a time. If I apply the eq settings from let’s say left to both left and right the imaging is great. ( the path length differences have been corrected using signal delay on the near side)

I’ve tryed a few write ups on rephase to correct the system and make it minimum phase and that part sounds great now I’m starting to get the hang of it a little.
It would be so fantastic if there were two windows in rew and rephase for each channel and software somewhere within one of them that would warn you when the system will have imaging issues because of the use of independent eq.

Auto eq makes a great 1ch filter. It needs to do left and right and output a set of corrections that won’t de-correlate. Or if anyone can tell me how to do this.

Super awesome super appreciated

Andrew

 

[jtalden]

I'm not positive, but think that aligning the L, R filter timing would be controlled within rePhase rather than REW. The REW suggested filters are all IIR type (minimum phase) so if implemented as such in rePhase they do not impact the relative timing of the channels even if the filters are different. In rePhase, where we have the option of FIR filters that could result in timing offset, we have the option to locate the impulse peak at a specific position in the 2 FIR filters. If we pick a position that is the same for both filters then the delay created will be the same for both filters. The relative timing between channels is thus maintained. The rePhase control to do that is 'centering' where the location of the impulse peak can be dictated in several different ways.

As an example, I use a single FIR rePhase filter for both channels just to remove overall phase rotation. While it is the same filter for both channels, I still want to specify the impulse peak location to efficiently utilize the number of taps. For the this type of filter the vast majority of the filter action takes place prior to the impulse peak so I locate the impulse at 85%. This avoids an excess number of taps after the peak that has no function. We can import the rePhase FIR filters into Audacity or other program to inspect and confirm that the peaks are properly located and that the active taps are not being clipped. When the filters are aligned the delay is the same and we can thus safely use our overall DSP delay account for the different distance between L and R.

Considering that you are only discussing L vs R it probably does not matter if the rePhase filter peaks are closely aligned or not as the R vs L delay is normally measured and set in the DSP after the filters are installed and active and thus any significant timing offset in the filter would thus be accounted for anyway.

Just my general understanding, but hopefully helpful. The rePhase forum or possibly others here can be more authoritative.

 

[Andrew Slater]

Thanks for the reply

Yeah I think maybe you missed missed what I was trying to say, I probably didn’t say it right


So I’ll first express my goal
Which is for both transfer functions to match
Which Dirac is excellent at accomplishing
With rephase it’s all manual.

In the room I’m in, the path lengths are large compared to how close you are to the source.
So left and right have largely different responses from the 1 seat in the room.
Using minimum phase eq on them separately results in time domain issues that de-correlate the center image. The eq applied has to be the same for left and right for that to not happen.

I have a fir for left and right but most of the fir is just minimum phase eq which is no different than regular peq.

When I look at the transfer functions for L and R I can make them have very similar looking responses it just requires very different eq on each to get there.

I’ve tryed using one eq setting for both that makes both look as similar as possible and that sounds good, very good, just not as good as Dirac can.

It would be great if rew had two eq windows for left and right and it could look at the responses and output a correction that won’t cause imaging problems. Or at least have two eq windows and would be able to let you make the choice, or something. Or I’m missing something and it’s already there and I just don’t know to do it.

Can I use different eq and just make the measured phase look the same after I apply the FDW and pull the excess phase back. Another-words, do the excess phase correction than remeasure L and R from the center of my head with 1 measurement each , remove time of flight, and make both sides measured phase look the same? Would that even work?

Thanks , :-)

 

[jtalden]

The rePhase writeups you referenced are detailed as to the process steps.

In terms of the impact of applying that process to a car application I have no experience. I would expect that any phase 'correction' portion of the filter is likely to very misleading. The numerous strong reflections in the vehicle results a very chaotic phase response. I don't think that rePhase can address that.

I see no reason not to try it out though to see if it helps. It sounds like you have had at least some success.

 

[Andrew Slater]

Thanks jtalden

I just gave it another try and think I’m starting to get a better hang of things.


I am starting to think I was on the right track using the FDW and excess phase.
I think I need to avoid that areas that are comb-filtering and work between them. If I do that I can get some seperate L and R eq in and its working much better. Maybe I’m trying to do too much.....

I’ve been using EQ and pulling the peaks all the way down to the bottom of the comb-filter dips to make response flat. I think I’m going to start all over and try again and this time ignore the big dips. The big dips are only at my seat and other dips are at other places. I only care about my seat but I’m starting to think just because the dip is at my seat doesn’t mean the energy isn’t there in the power response.

I appreciate your help, still think two windows or a 2nd channel of some sort would be better if at all possible.

It’s starting to sound a lot like my Dirac tunes, I’m not quite there yet, almost

 

[jtalden]

Well done.
You are correct in using a heavy FDW to do this. There can be no benefit to adjusting phase disruptions due to reflections.