Premium Membership - All Voxengo Plugins at a Fixed One-Time Fee - Click Here For More Info
Forums     Plugins     CurveEQ CurveEQ matching algorithm & parameters

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.

Hi Aleksey,

If you ever had the mind to improve (?), that's not the right word...augment the matching algorithm in CurveEQ I thought I'd start this thread in case you were interested in some input from CurveEQ users.

I'm hoping there will be both pro & con comments in here as well as usage info that might improve the alogithm - or prove that it is just as it should be.  Either way !

Here's my 2 cents then...

CurveEQ matching usage catagories:

1.  Full mixes

2.  Single tracks of a mix

My current EQ matching method:

I use CurveEQ matching on full mixes and try to push the curve of my mix into the curve of a professional mix a bit (not all the way of course).  It seems to me that CurveEQ matching usually wants to change the bass and highs way too much so I have to manually adjust how much I allow CurveEQ to change things.  One way to control CurveEQ is to choose less points to match - I usually use 12-30 points with the current matching scheme.  Still I may want to change the bass and highs even more so I will use the attenuate buttons to change the new curve to 0dB of change, flat in other words.  Then I will select the bass frequency range (20-300Hz) and amplificate it to taste (I use my updated monitoring system to hear when it sounds correct - computer speakers won't do using this method).  I do the same thing using the mid and high freq ranges.

My Wish List for matching EQ(brief):

What I think would help me is more resolution, maybe an FFT of 8192 (I'm not sure this is correct - just guessing) because I seem to need more detail in the bass when matching where a lot of resonances in home mixes exist.  If CurveEQ had more bass resolution then I would ask for more matching points (currently max=60) to better define the bass resonances I see.  With more points you don't necessarily want to use all of them equally spaced across the spectrum so maybe a parameter would help that would 'cluster' a higher percentage of the points in the bass region, somewhat less in the mids, and some appropriate amount in the highs.  Maybe this could be selectable by the user - maybe just using a logarithmic distribution might work too.

Another possibility might be to interpolate some degree of matching starting at the bass end of the spectrum and gradually adjusting in percentage depending on some user selectable slope to the high end.  I believe that I would normally use more matching in the bass and mids than I would in the highs so I might select a -3dB/octave slope for starters.

I think you see what I'm getting at here Aleksey, I don't know if there's any bang for the buck messing with the matching algorithm but some of us were talking about it over at the musicplayer forums and I thought I'd ask.

You have so many other algorithms that might also be useful to matching such as the GlissEQ dynamic and the Transmodder algos that there are a lot of choices.  I mention this because the common thought in matching is that if there isn't anything happening in the freq range you're matching with a gain in dB then the noise floor or some other useless audio information would be increased.

I think matching is important for the home recording enthusiast that has minimal monitoring and experience like me.  After producing a home DVD out on the road here (I'm in Knoxville, Tenn wrapping up some business) using consumer tools like Media Creator, Magix and some other things I can tell you that there is a new market of home enthusiasts out here.  DVD home producers are going to need tools that create acceptable audio using only 2" laptop monitors and a set of Beyer DT770 headphones.  Out on the road the headsets sound better than anything else practical to carry.  That tells me I need some good reference curves and matching tools to get the EQ corect.

What do you think ?  Do you have time for that kind of thing ?

Thanks for sharing with your ideas.  I'm currently collecting various ideas/approaches for the possible upcoming CurveEQ 3.0.  I'm going to use every single technology I have so far and probably ideas you have mentioned can be implemented.  But please bear in mind that I do not have any pre-set dates for the release.

OK - Thanks for considering that stuff for some future release Aleksey!


This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.