Premium Membership - All Voxengo Plugins at a Fixed One-Time Fee - Click Here For More Info
Forums     Plugins     CurveEQ Rebalancing a mix spectrum ??

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 and CurveEQ users,

I'm looking for some advice concerning using CurveEQ to rebalance the EQ of a mix that has basic EQ flaws.  In other words it's a little early to consider sheen, polish and other 'mastering' type suble enhancements - I'm talking EQ correction here.

Given that the 'curve' or 'spectrum' of both my reference mixes and source mixes needing rebalance have flat areas, peaks, and valleys I am trying to use that visual information to try and preserve the 'good EQ', get rid of the 'bad EQ' and possibly bring out the 'hidden' good stuff (that's the polish and sheen piece).

Comparing the average/peak spectrums between my source material and reference material (similiar in nature, 4-5 piece rock band, similiar tempo, and room) the peaks and valleys and flat parts of the curve sometimes line up and sometimes not.  This is expected since I'm looking at 2 different (although similiar) pieces of music that have different frequencies emphasis because of different instruments, signal chains, rooms, vocals, etc.

What I'm thinking is that I don't necessarily want to push a source band with a valley up to match a reference peak band just because the reference has a peak at that point.  There may not be the right kind of musical energy in the source band I'm pushing up.  Instead of pushing a narrow valley up in gain I think I might want to take a wider view of that area and open it up to include more 1/6 octave bands.

What does spectumatch do in this situation ?  How does it handle valleys in the source or does it even have enough information to even worry about it ?  Are valleys, peaks, and flat areas equally weighted and matched no matter how big the delta ?  I think the answer is yes.

I guess I can't really say that just because there is a narrow valley there is not good musical information in there, but I also can't say that it should arbitrarily be pushed up to match a reference.  This is where the ears come in I guess.  I'm using spectrumatch to give me a boost in certain cases because at this stage of the game for me I'm not too fast a rebalancing a full mix.  So I'm learning to attenuate the points I don't want to rebalance and then select the points I do want to rebalance and use the attenuate/gain arrows to taste.  In that respect then I'm using spectrumatch to automatically give me 'potential' points for adjustment - my ear may not like them so I can use the arrows to suit.

I have noticed one thing - sometimes after using spectrumatch If I attenuate back to 0db gain then it's a little easier rasing the gain a little bit at a time to find the correct gain.  Sometimes the problem with loudness is - if you start at the loudest point then you attenuate things may sound not as good the quieter they get due to the ear sensitivity - another variable.

Well, any tips or comments on spectral balancing would be appreciated.

Traditionally mastering engineers have used wide bandwidth adjustments -- so obviously there is a lot that can be done with such a technique (i.e. everything).  I would have that you would be best off not worrying about narrow peaks and troughs.  I almost always set set the meter resolution to 1/6 or 1/3 octave resolution -- I think it gives a clearer indication of what I actually hear.  However, I have never yet had a successful result using spectrumatch, so I prefer to eq manually, using the average meter to give a reasonably straight line over the most audible part of the spectrum (I use 5dB octave slope).  I'm no mastering engineer though (I find mixing much more straightforward), so I'm always looking for tools to make it easier.  I'd also love to hear how other people use CEQ for mastering.

Yes, Andrew's approach sounds better to me (getting a 'straight' line at 5 dB/oct slope, manually).

I believe SpectruMatch can give 'sane' results with 30 or less destination points.  SpectruMatch cannot be an universal solution.  It is good to 'give a direction' towards some particular EQ balance.

The only time I have tried to use CurveEQ in mastering was to add color through "vintageness," saturation or gear match.

I have never had much luck with frequency matching programs like FreeFilter or Spectrumatch, because all the other variables in recording techniques, dynamics, player's instruments, etc. prevent, for me, a usable "match."

Like Andrew, I use my ear and manual EQ, the best I can.  I'm not very good at remixing anyway, so I don't really try.  If I find I have to use too much EQ to get a decent sound, then I try to have the song remixed.


OK Andrew, great tip for rebalancing my mix !

I set the the slope to 5dB/octave (I was using 3dB), 1/4 octave resolution (like the Avalon I was reading about today !), Average mode, drew a freehand curve around the stuff over the 0dB line, inverted the filter - bang zoom.  Perfection in less than 5 minutes (3 minutes was because I never used freehand mode before).

This is absolutely the closest balance I have gotten yet !  Bass mids and highs - smooth and full and shiney !  The sound was in there the whole time...

Thanks Andrew!

You've heard Joe Meeks saying - If it sounds good, it is good.  I think Aleksey gives us the tools to say - It sounds easy, because it is easy !

Good tip, thanks again Andrew.  Great tool Aleksey !


PS kylen says - If you make it hard, it is hard !  Ha Ha - time for dinner in California !

Ed.  17SEP2003 Well I can't get a good sound using this slope now on some other material so it just depends I guess.  I'm back to the +3dB/octave slope and viewing the entire freq range 20-22.5K, not just the midrange.  I've also got an Eqium EQ ahead of CurveEQ to give me about 4 or 5 narrow band cuts (.06-.08 octave) where there are some really bad peaks and using a multiband compressor makes it sound worse.  Supposedly it is fairly hard for the ear to detect a narrow EQ cut of 1/10 octave or less - I guess that's what seems to be true also.  Spectrumatch seems to work better this way too if it has a better balanced wave to listen to - I still have to attenuate the match quite a bit (keep points within +/-2dB) to make it sound good. :)

One observation here.  In the past records had -6 dB/oct rolloff.  Today the norm is -3 dB/oct (much brighter).

The other thing is that you don't have to get a 'straight line' response.  I've noticed that many commercial recordings have a (wide) 'curved' response peaking at around 600-700 Hz.

So, after getting a straight line response you can additionally EQ with that curved response and you'll eventually get that 'compatible' commercial sound.

One more note: you don't have to get a 'good' sound when first balancing the material to get a straight line response.  You can finally adjust the sound when applying the curved EQ.

Strangely, I've just tested this approach and it worked very well...  I've used OutPeak mode to got a nearly straight line response and then applied that curved EQ (very simple to do it with CurveEQ).

Of course, you have to choose spectrum slope first.  Choosing an unsuitable slope will result in a too bright/too dark mix.

Yo Kylen :-) I had never thought of using the free hand mode in this way (balancing around the zero line, and inverting) -- good tip.

Aleksey -- since you have obviously analysed the spectral patterns of music -- how about providing some target curves along with CEQ.

In my experience, of looking at other mastered stuff, the average bass level usually starts to roll off (i.e. drop below the line) below 80-150Hz, and treble drops off above 12-15KHz -- these figures are done from memory so are pretty approximate.  It seems to me that the linear part of the spectrum of a well mastered cd really lies between the high bass and low treble, the low bass and high treble are pretty variable in how much they roll off and at exactly which frequencies.

Aleksey: "One more note: you don't have to get a 'good' sound when first balancingthe material to get a straight line response.  You can finally adjust the sound when applying the curved EQ."

Sorry, I don't understand what you are saying here.

From your last post, I'm interested -- why did you choose outpeak rather than outave mode?

Come on you masters -- let us know how you use CEQ :-)

Providing 'standard' curves is not a good way I guess.  I think everyone can do that kind of mix balancing.

Yes, you've got the correct idea about the rolloff, but my observations say the rolloff is pretty constant (forming a bell curve with a mostly flat top with side rolloffs).  I.e. top of the curve is not flat, but also has some rolloff.

Speaking about the 'good' sound, I've meant that you don't necessary get the good sound when forcing the mix to follow a straight horizontal line (any sound flaws must not be taken into account on this step).

All faults must be taken into account on the step of building that 'bell' curve.

I've also used OutAvg mode, but I guess OutPeak works good for this kind of rebalancing, too.

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.