Login / Register
 
Theoretical Question on EQ subtracting/adding
New Reply
Subscribe
Edmond Dantes
Thread Starter
#1
11th May 2007
Old 11th May 2007
  #1
Gear nut
 
Edmond Dantes's Avatar
 
Joined: May 2007
Posts: 100

Thread Starter
Edmond Dantes is offline
Theoretical Question on EQ subtracting/adding

Hey everyone. First post.
I come from the realm of composing so chopping my way through production techniques and asking a million questions on the way.

So i was just thinking about EQ. I want to know if this would be the same thing.

I record a guitar (or whatever) naked ( no effects etc..) and somehow record the exact same thing being EQ'd ( say a small curve around 440 at -2db). Now if I take the wav file of that recorded guitar, with the negative EQ curve, import it into pro tools and play it setting it up with the opposite EQ (exact same curve at same frequency at +2db) will I have the same as the original audio? In otherwords can that compensating +2db boost produce the same sound as the original or is something lost in the signal chain by never recording those frequencies at all?

Thanks, i am interested to hear what you guys have to say.
#2
11th May 2007
Old 11th May 2007
  #2
Lives for gear
 
Watersound's Avatar
 
Joined: Jan 2005
Location: New Jersey
Posts: 721

Watersound is offline
I think the first answer is why would you do this if you already knew you would have to compensate for the eq cut later on? I suppose you can restore that curve but every eq sounds different so it would never be identical to the original, untouched track. But I doubt the difference would be very dramatic in most cases.
__________________
JD
#3
11th May 2007
Old 11th May 2007
  #3
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
This sounds too well phrased to not be a test...

But I'll bite just to get things going.

I believe that getting the "exact same curve" would be rather difficult, but not impossible. Each EQ's has it's own characteristics: a Q of 2 isn't exactly the same for every EQ, etc, and each EQ has it's own harmonic distortion, even in the digital realm, etc.

If you were able to exactly match those type of things for the situation you mentioned, I'd say you'd be able do a null test between the EQ'ed then un'EQ'd version and the completly non-eq'ed version and get a null into the noise floor.
__________________
<X>
www.noodlinstudio.org
Edmond Dantes
Thread Starter
#4
11th May 2007
Old 11th May 2007
  #4
Gear nut
 
Edmond Dantes's Avatar
 
Joined: May 2007
Posts: 100

Thread Starter
Edmond Dantes is offline
Let me assure you this is no test, just a genuine curiousity I have which can have so many possible applications. For one example regarding high frequencies:

Say one was to record a low to mid passage on a piano. Say due to a hypothetical budget problem these microphones (or singular mic) had a hard time recording the highs of the piano then say a more expensive mic. Now coming from the acoustic world, of which I am most familar, I would have to say that no EQ would be able to compensate for seemingly weak hi's exactly. Why? Well according to what I know so far - the overtone series. Would drawing in an EQ curve to fix this problem work? I don't see how it can. But does that mean there would be an audible difference between the orginal and the latter? Will the EQ'd passage sound artificial to our ears?

This is an important matter because I work in samples all day. Some, to a majority of the, samples have great sounding mids and weak highs. It is a chronic problem. Now I ask this question. If the highs are not there on the original sample will any amount of EQ'ing get me to the point where the "non weak" or "acoustic" sample was Or does the overtone series and other factors I don't know about not allow this to be done? I find this to be very confusing.
#5
11th May 2007
Old 11th May 2007
  #5
Gear Guru
 
drBill's Avatar
 
Joined: Jul 2006
Location: So Cal
Posts: 15,076

drBill is offline
Why don't you just try it??

Most digital EQ plug-in's can be very exacting. Much more so than their analog brothers. BUT, there is a difference between cutting and boosting. When you boost, the phase tends to get more whacked out than when you cut. It's been a well known technique for years. As a matter of fact, some engineers rarely ever use the boost of the EQ, prefering instead to cut - because of what boosting does to a sound. Now, I'm not a purist, so I'd suggest this:

1. record a MONO guitar, piano, etc., into you DAW
2.) Use a EQ plugin to do a boost - let's say + 2dB @400Hz with a Q of .7
3.) use an internal buss to record the EQ'd track to a new track

NOW,

4. Cut 2dB @ 400Hz with a Q of .7 on the original track and again internally bus and record to a new track.
5. Reverse the phase of one of the rendered EQ'd files.
6. Pan the two tracks mono and bring them up to unity gain. (One is phased reversed.)
7. If the EQ is consistant between cutting and boosting, you will near nothing. Complete silence.

My guess is that it will not completely phase cancel out to nothing, but that the two sounds will sound pretty similar. Keep in mind that good digital EQ doesn't phase smear as much as decent analog EQ. That's kind of why we like analog EQ - the way it screws up the sound. Kind of like why we like tape - same reason. It changes sound in a way that has become pleasing to out ears.

8. Most important step - report your findings back on this thread. now get to it!!!!

bp
#6
11th May 2007
Old 11th May 2007
  #6
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
We'll, this topic definitely spurs my curiosity, too.

I think you have to start from a theoretical perfect model, which is how all engineering of electronics, or anything scientific I know of for that matter, starts. After understanding that theoretical perfect model, you start adding in the imperfections that reality gives us into the results. As you start adding in those real life imperfections, the questions that you are asking become answered.

If there were no imperfections in the signal chain or in the recording process itself, then getting the signal back to what you started, shouldn't be a problem.


You mentioned something that we really have to observe as important here: recording the exact same performance through the exact same mic and preamp, and at that point putting only one signal to an EQ right before the 2 are converted to digital (this is more precise than going to tape and repeatable between the two tracks even if not perfect).

Is this the scenario?
#7
11th May 2007
Old 11th May 2007
  #7
Lives for gear
 
Joined: Feb 2006
Posts: 4,075

Kiwiburger is offline
For a start - you can't boost what ain't there. If your source is lacking in high end, attempting to boost the highs will only boost the noise floor.

If you completely cut some frequencies with your first eq curve, then you could not restore them with an inverted eq curve. If you just applied a small gain reduction to a band of frequencies, then you could restore them with a small boost. But you would also be boosting the noise floor.

EQ is destructive. So don't use more stages than you need to.

Except some certain linear phase eq's, all eq's shift some frequencies more than others in the time domain. Inductors and capacitors take time to charge.

From that perspective - eq'ing twice will double the phase shift problems. Although if you are extremely clever, you can reverse the audio the second time - and the phase shifts will then work backwards in time, cancelling the first shifts out. Then reverse the audio and you have a linear phase eq effect ...

With eq - if in doubt, don't.
__________________
My carbon footprint is bigger than yours.
#8
11th May 2007
Old 11th May 2007
  #8
Lives for gear
 
swankdoc's Avatar
 
Joined: Jun 2005
Posts: 2,138

swankdoc is offline
right, I was going to reply like the downunderbergeronthebarabi.

If you record a source with a cut EQ, then certain frequencies / harmonics and basic wave interactions are also cut. They're just no longer there (to some degree depending on how hard you hit it.) and it has no way of knowing what could have been there. Its now its own sound, lacking a certain frequncy dip for sure, but other stuff like reflections, nodes, peaks etc.
So you go and try to rebuild this floor with a new EQ boost, and you're just boosting what new floor you made. I cant see how that could ever approach the original . It all depending on how crazy you get.

Say you take it to the extreme, and whack all 4 bands of a parametric EQ by -10db or so. Lots of low lying mud all over the place. THere's no magic water booster EQ in the world (thats not for drinking) that would be able to somehow magically recreate was was there before. Its just gonna boost the new muddy signal.


Thats my story and Im sticking someone else to it.
#9
11th May 2007
Old 11th May 2007
  #9
Gear addict
 
dropblacksky's Avatar
 
Joined: Sep 2006
Location: San Francisco, CA
Posts: 495

dropblacksky is offline
Just wanted to say that I was about to post a naysayer response to the idea that the eq could recreate the original tone...that phase relationships would be changed and that an eq can't create what isn't already there..blah blah blah.....then I tried it.

Recorded two tracks similtaneously in pro tools...one straight, another through an aux track with the original poster's EQ of -2dB at 440Hz. Lined up the two tracks at sample level, added a 2dB boost with the same eq plug on the original eq'd track. Threw one of them out of phase...and BAM...Null cancellation....silence. Kinda makes me wonder about how much of a commitment eq'ing to tape really is....
#10
11th May 2007
Old 11th May 2007
  #10
Lives for gear
 
Watersound's Avatar
 
Joined: Jan 2005
Location: New Jersey
Posts: 721

Watersound is offline
Quote:
Originally Posted by Edmond Dantes View Post
Now I ask this question. If the highs are not there on the original sample will any amount of EQ'ing get me to the point where the "non weak" or "acoustic" sample was Or does the overtone series and other factors I don't know about not allow this to be done? I find this to be very confusing.
No, that is why you always hear people say "get it right on the way in." If it were that easy think about how much better we could make lame, thin vocals sound, etc. Let's always remember that with this type of approach we are in turd polishing territory. Shiny turds are still turds, so either get better samples or use real musicians/good instruments/mics where you can control the harmonic content going to tape/DAW.
#11
11th May 2007
Old 11th May 2007
  #11
MonsterIsland.com
 
Joined: Sep 2005
Location: New York City
Posts: 4,496

Mike Caffrey is offline
Cool to hear someone tested it.

I htink it's an interesting theoretical question and I think theortically, if the optimal condistion were met, it should be possible to make the same sound. I don't really understand how this changes anything when working with samples.


I good example of this principle is Dolby Noise Reduction. Dolby prints to tape with a pre-emphasis in teh high end and then de-emphasizes the high end on playback to give you the same signal you started with, but it's de-emphasized the hiss.

I'm the sound of Dolby NR is wll enough known that people can say whether it's accurate or not. There is some compression as well, so it's maybe not a true test.
#12
11th May 2007
Old 11th May 2007
  #12
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
Quote:
Originally Posted by dropblacksky View Post
Recorded two tracks similtaneously in pro tools...one straight, another through an aux track with the original poster's EQ of -2dB at 440Hz. Lined up the two tracks at sample level, added a 2dB boost with the same eq plug on the original eq'd track. Threw one of them out of phase...and BAM...Null cancellation....silence.
Cool. Thanks for doing this. I'll have to try this when I get a chance, but A few questions:

1) What was used for the recorded EQ (the -2dB at 440Hz): a physical mixer or a plugin?

2) Did you use a bit meter or a spectrum analyser to see how much they cancels?

Now the interesting thing would be to use 2 very different types of mics as close to the same position and see how closely they cancel. If the system is clean enough you could use this to show differences between mics.
#13
11th May 2007
Old 11th May 2007
  #13
Gear addict
 
dropblacksky's Avatar
 
Joined: Sep 2006
Location: San Francisco, CA
Posts: 495

dropblacksky is offline
Quote:
Originally Posted by NoodlinXavier View Post
Cool. Thanks for doing this. I'll have to try this when I get a chance, but A few questions:

1) What was used for the recorded EQ (the -2dB at 440Hz): a physical mixer or a plugin?

2) Did you use a bit meter or a spectrum analyser to see how much they cancels?

Now the interesting thing would be to use 2 very different types of mics as close to the same position and see how closely they cancel. If the system is clean enough you could use this to show differences between mics.
My test was definitely more practical than scientific. I used a plug in because that's all I have at home. Perhaps "null cancellation" is an extreme term since I didn't use any tools besides my ears for the test. I did, however, turn my monitors WAY up after flipping the phase and still heard nothing. It would be interesting to try it with outboard gear, or maybe harware going in and a plug in emulation of the hardware once in thte daw...see how close the emulation really is.
Edmond Dantes
Thread Starter
#14
11th May 2007
Old 11th May 2007
  #14
Gear nut
 
Edmond Dantes's Avatar
 
Joined: May 2007
Posts: 100

Thread Starter
Edmond Dantes is offline
I agree with the idea about, you can't boost what's not there. Perhaps the original example of -2 DB is too small to make a difference. But this leads me to another question. However it does seem to make sense that you can decrease in signal what is there. Because you are just lowering it to a smaller level.



What makes a mic a mic? Again I am a novice.

1) Spacial pattern. Figure 8 etc...

2) Size/function - Large, pencil, ribbon

3) is the rest EQ?

*There must be someting else? Or else why wouldn't companies just make a mic that records alot of everything and then sell very specific EQ patterns to subtract what was needed to duplicate say a brilliant sounding U47 from the golden days. All of these questions tie together from my original one. Which i still kind of don't get very well.
#15
11th May 2007
Old 11th May 2007
  #15
Lives for gear
 
soundbarnfool's Avatar
 
Joined: Jan 2007
Location: Nashville
Posts: 887

soundbarnfool is offline
You ought to talk to some mastering guys about this. I would say that whenever you put any signal through a box of any kind it changes character. Also, lows act differently from highs, etc. When recording dialog in the field, I always cut 100Hz to get rid of rumble, etc., and it is gone for good. You couldn't get it back by adding 100Hz in post: what you'd get would be bassy dialog w/ a bit of the original rumble peaking through perhaps. Or perhaps I should say that you couldn't get the rumble back in it's original incarnation (pre-cut), because the total sound picture has been drastically altered by the cut: the ratio of dialog to rumble has been increased. I would imagine these principles apply to recording music as well, but I use EQ as a last resort and work for days on mike placement and tweaking performances so that I end up using very little EQ.

Good thread.
#16
11th May 2007
Old 11th May 2007
  #16
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
The sonic color or character of equipment can be defined and measured. It really comes down to 2 things: time and amplitude. If this wasn't true digital wouldn't work so well. And we wouldn't have convolution.

Quote:
Originally Posted by Edmond Dantes View Post
What makes a mic a mic?
Frequency response (EQ)
Polar pattern (where you aim the mic)
Distortion charactistics (harmoic, phase, amplitude)
SPL handling (how loud is your sound?)

However, you can't stop there because you can change any one of those things and it affects the others. Put the mic in front of a louder sound and the distortion characteristics change, put the mic closer or to the side of the sound and the frequency response changes. And we haven't even introduced the room response into this...

To start with it's a very simple set of parameters to deal with and understand, but when they get put all together in the real world the result is a very complex system of parameters that are interelated.
#17
11th May 2007
Old 11th May 2007
  #17
Lives for gear
 
max cooper's Avatar
 
Joined: Aug 2004
Location: tx
Posts: 8,802

max cooper is offline
Quote:
Originally Posted by Kiwiburger View Post
For a start - you can't boost what ain't there. If your source is lacking in high end, attempting to boost the highs will only boost the noise floor.
The RIAA de-emphasis and re-emphasis works kinda like this, right?

And just because a source is lacking in high end doesn't mean it's not there, it may just mean that there's less of it.

And I don't see any reason you'd only be boosting the noise floor. Wouldn't, for example, a high shelf boost everything it *sees*?

RIAA equalization - Wikipedia, the free encyclopedia
#18
11th May 2007
Old 11th May 2007
  #18
Gear Guru
 
drBill's Avatar
 
Joined: Jul 2006
Location: So Cal
Posts: 15,076

drBill is offline
dropblacksky - thanks for doing the test. When I've got some time I want to try it with different types of plug-in EQ's. I suspected that it would be very close to cancelling, but completely silent???? That's pretty impressive for the plugin. Still it's all how we "hear" it ultimately. In the land of analog EQ's you wouldn't have gotten silence, maybe close, but the tolerances between electrical components just can't be as precise as doing it in the digital world. Again, thanks for taking my test! You get an A+

thumbsup

Bill
Edmond Dantes
Thread Starter
#19
11th May 2007
Old 11th May 2007
  #19
Gear nut
 
Edmond Dantes's Avatar
 
Joined: May 2007
Posts: 100

Thread Starter
Edmond Dantes is offline
But it I did a high pass from say 440 again. Then imported that file and boosted a huge DB boost below 440 and then made it normal at 440 up it should be the exact same as the high pass file. Better yet if I muted above 440 and pumped +10 DB below and played the imported file i Should still get silence? If not what then am I hearing?
#20
11th May 2007
Old 11th May 2007
  #20
Lives for gear
 
kafka's Avatar
 
Joined: Apr 2007
Location: Maryland
Posts: 6,405

kafka is offline
Quote:
Originally Posted by drBill View Post
dropblacksky - thanks for doing the test. When I've got some time I want to try it with different types of plug-in EQ's. I suspected that it would be very close to cancelling, but completely silent???? That's pretty impressive for the plugin. Still it's all how we "hear" it ultimately. In the land of analog EQ's you wouldn't have gotten silence, maybe close, but the tolerances between electrical components just can't be as precise as doing it in the digital world. Again, thanks for taking my test! You get an A+

Bill
Yeah, I was able to repeat the test. Dead silence. Moreover, when I summed them to another track, it read zero. That's not really that surprising, now, is it? ITB it's just math, so we should have complete repeatability. What's more interesting to me is exactly the test you propose with analog equipment.
#21
11th May 2007
Old 11th May 2007
  #21
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
I pose a few questions that come to mind when thinking about doing a null tests. I believe it's completely on topic:

If dithering only affects the lowest bits, which at the noise floor level, how can it make a noticable difference to the sound we hear?

Does this mean that the sonic color a device or process adds to a signal only needs to add a minute measurable difference for us to be able to hear it? Furthermore, if we nulled the difference out would we be able to hear it?
#22
12th May 2007
Old 12th May 2007
  #22
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
OK. I tried this with the EQ on my hard mixer with poor results. I feel it's rather inconclusive however possibly useful information so I'll report it. Maybe someone will point me to an error?

Otherwise, I might need to try better cabeling, different channels, or just a better mixer. This does show a good way to see if the channels on your mixer perform the same.

Unluckily, I don't have an EQ bypass switch on the hard mixer, a XLR Y cord, nor a 1/4” Y cord (where did the Y's go?). So I used some adapters and a stereo RCA cable to plug a guitar into two channels on the mixer.


I wanted to make sure I could NULL out the difference between the two hard mixer channels with no EQ on either one, and I couldn't. I zeroed the EQ, inverted channel 2, and while playing I adjusted the volume of one of the channels until I got the most cancellation. Then I used that for the experiment:


Recorded into a DAW. All measurements from Voxengo Span. In recording 2, Channel 2 software EQ adjusted to find best cancellation at 352hz, Q 1.5 octaves, +7.5dB. Although, I didn't play with the Q.


Recording 1: Both channels flat EQ
Normal: -16.2 dBFS Peak RMS
#2 inverted: -68.8 dBFS Peak RMS


Recording 2; Channel 2 with hardware EQ: -6dB at ~400Hz Q = 1.5 octaves
Both channels: -11.8 dBFS Peak RMS
Both channels, Channel 2 inverted: -20.4 dBFS Peak RMS
Both channels, Channel 2 inverted plus soft EQ boost: -25.3 dBFS Peak RMS
#23
12th May 2007
Old 12th May 2007
  #23
Gear Guru
 
drBill's Avatar
 
Joined: Jul 2006
Location: So Cal
Posts: 15,076

drBill is offline
Quote:
Originally Posted by NoodlinXavier View Post
OK. I tried this with the EQ on my hard mixer with poor results.
As I said before, there's no way to get that to happen on a hardware mixer. You're dealing with electronics that have different specs, where there are resistors and other components that (even if they use 1% tolerance) will not match up enought to completely null out the sound. You would realistically also need stepped pots that were very closely calibrated - as on a mastering EQ. Also, boosting smears the phase more than cutting so.....as it sounds like you found out, re-creating an "un-EQ'd" sound by reversing the boost or cut is not so easily achieved.
#24
12th May 2007
Old 12th May 2007
  #24
Gear nut
 
NoodlinXavier's Avatar
 
Joined: Nov 2006
Location: Minneapolis, Mn, USA
Posts: 87

NoodlinXavier is offline
Quote:
Originally Posted by drBill View Post
As I said before, there's no way to get that to happen on a hardware mixer. You're dealing with electronics that have different specs, where there are resistors and other components that (even if they use 1% tolerance) will not match up enought to completely null out the sound. You would realistically also need stepped pots that were very closely calibrated - as on a mastering EQ. Also, boosting smears the phase more than cutting so.....as it sounds like you found out, re-creating an "un-EQ'd" sound by reversing the boost or cut is not so easily achieved.

I do electronics for a living and agree with what you say. However, I expected to be able to cancel much more than I did: 52 dB PRMS when both channels are flat?

If I had the ability to bypass the EQ sections, would I be able to get more cancelation?
Edmond Dantes
Thread Starter
#25
12th May 2007
Old 12th May 2007
  #25
Gear nut
 
Edmond Dantes's Avatar
 
Joined: May 2007
Posts: 100

Thread Starter
Edmond Dantes is offline
So in theory if these signals are supposed to cancel out by phasing wouldn't what is left be what sound the machine (or signal chain) itself produces? If true, it could be an interesting way to find what a piece of gear adds to the signal chain.
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+ 
 
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
blackcom / So much gear, so little time!
7
electric / So much gear, so little time!
3
landshark / High end
3
exfakto / So much gear, so little time!
9

Forum Jump

SEO by vBSEO ©2011, Crawlability, Inc.