I have figured out a way of tuning my EQ which I would like to share with you all.
Some of you might not like it but it is working for me and I am sure this will help some people.

The music I listen to is electronic such as house and hi-nrg but it should work for other genres.

What you will need is a good audio source(such as computer with a soundblaster card) and a pair of decent headphones to use with it, this is to use as a reference.

I have a 13 band EQ in my car but most MP3 players are only 10band, so I got a plugin for winamp so it could have a 13 band, the plugin also has a 31 band.

I done the EQ tuning 1 frequency at a time and the very high and very low frequencies are the easiest to do.

Now lets say you want to tune 400Hz.
Go through a few songs quickly on your reference source with everything flat except for the frequency you want to tune, have that almost all the way up at around +9dB. Keep listening and you will eventually get to an instrument or sample that it will drastically change the sound of, if you are not sure go over it with that frequency all the way down, if you were correct it should be barely audible. In my case boosting 400Hz made some basslines more characteristic and less of a generic "donk". I then played that same part of the song in my car fiddling with 400Hz to get it where I thought was a good balance of having the character of the bassline audible but not overpowering the actual bass.

The same goes for other frequencies, the trick is to find a particular part of a song or 2 that is changed most by the change in the frequency, and to keep going over it, you can do this without another audio source but headphones have the advantage of superior clarity so its much more easier to find out what the ranges are actually doing to your music.

There's an even simpler way to tune your system if you don't have access to an RTA. Use a pink noise track and begin by setting all the EQ flat. Forget the low bass as this is best tuned by ear with music and just focus on the region from 100Hz and up.

Since the ultimate limitation to tuning the type of EQ processing itself all you can do with most systems is pick a tuning band and - at moderate volumes - apply boost and cut one band at a time and listen to the pink noise itself. It will be obvious to your ears that somewhere on the boost/cut range of each band the pink noise will sound 'smoother'. It helps to close your eyes when doing this by the way - kind of like it helps to turn the stereo down when you do a reverse park.

Anyway, go through each band with this routine and by the time you've done all of them the pink noise will sound nice and even across all the audio spectrums, and you'll be quite surprised to find that the resulting response will be quite smooth and flat - or at least as flat as th EQ limitations allow.

I've done this a few times for mates cars when I haven't had an RTA at hand.

BTW, note that there are widely varying EQ types to tune with. Graphic, parametric, quasi-parametric and so on. And with either type there the Q factor to consider and - with graphic - if you have constant Q so that varying neighbouring bands does not affect those adjacent.

Sadly, EQing isn't such a simple task unless you've got a ton of experience under your belt.

My system had already been tuned with an RTA to have an almost flat response, and that's exactly how it sounded, flat.

Interestingly, with the RTA tune 630Hz was max at +9dB and 400 was at like -1db. Listening to my music I have 400Hz at +5 and 630Hz at +3. big difference but it has balanced out the sound of my music a lot.

The only problem I am having is with the 1-4kHz ranges, I think I am going to leave them flat.

it will alway sound awfull tunninig to a perfectly flat response across the Hz range. when it is said to have a flat response it is accepted to have the hz next to each other on 1/3 octave eq, can be +/- 3db.. And generally sounds better starting high in the bass region and gently sloping downward towards the highs but its going to depend on alot of things

Example: 32hz-0db, 40hz-+3db, 50hz-+6db, 63hz-+3db, 80hz-0db, 100hz--3db

That is said to be flat even though there is increases and decreases

Interesting tuning method. But I reckon tuning with pink noise or a well recorded CD will probably still be more popular. What you're forgetting is that you're tuning to one particular CD, and one genre. You tune for that, other genres are not gonna sound that good. Dont spose it matters if you only listen to one sort of music though

I got good results from the Rives Audio Test CD and a Radio Shack SPL meter:



Apparently the meter is very repeatable, but lacks a nice flat response. The Rives Audio Test CD has a modified response to accurately counter the poor response in the Radio Shack meter, theoretically giving you a nice flat playback response when read with the meter..

Probably one of the biggest myths in Sound reproduction.

Flat does not sound bad, it just sounds bad on 90% + systems because most peoples systems lack Dynamics aka amplifier headroom and cone area.

Go listen to set of mini monitors setup for flat reproduction being powered by barely enough power, add some more power and it will sound alot better. Now go find some mega sized monitors with loads of power behind them and then tell me flat sounds bad.

my experience is that ppl simply do not know how an EQ works. they dont realise that by using an EQ (unless its something like a DEQX) they are applying a filter which means that they are also altering the response in the time domain as well.

so they blame the EQ when they have a +9dB boost instead of using the tool properly.

its like using a hammer to belt a screw into a bit of timber instead of using a screwdriver. wrong tool for the job.

Agreed. Flat isnt bad at all, flat is in fact the way you should set it up. It means that the reproductive system is neither adding nor subtracting anything from the recording.

Unfortunately, unless you have linear-phase filters, both EQ AND Xovers, you're going to find it hard to tune flat without impacting both on phase response and impulse.

raises the question then - is impulse / phase response more important than amplitude linearity?

one cant be at the expense of the other

its about finding a balance between the two. that is the 'art' of sound reproduction. thats where experience comes into the equation.

floyd toole has a lot to say about amplitude linearity

What do you mean by this?

And yea Luke, very true. I was just listening to the Scarface soundtrack and although it sounded really good, I could easily tell it would of sounded better on the original flat tune. But I only really listen to one type of music and I am very happy with the way I have made it sound.

The area between 1kHz and 4kHz is baffling me, its as though it doesn't matter if its afew dB off, where as with 400 and 630, I think I got it perfect for my liking.

Almost, but IMO, not entirely. Big problem is phase. Extremely difficult to measure, quite difficult to adjust how you want or correct a change, also difficult to audibly hear changes until you do something drastic. Impulse response is almost always adjusted mainly through install too (Q mainly). So in the end, you have three very important factors, but one of them is easy to measure, easy to change, and fully reversible. Guess which one jumps into the 'tuning' bag of enthusiasts....

There are two domains that coexist. One is the time domain, and when measured, can show an impulse response (feeding a 'snap' signal -on/off, into a unit and measuring its amplitude over time, or time to rest) which can give us an incredible amount of information about damping, strength of motor, environmental responses (like reflections) amongst other things.

The other domain is the frequency domain (also inclusive of phase) which shows amplitude over frequency (you've seen 1/2 of it as frequency response. The other half is phase over frequency). This gives us other information, all very important.

See, the thing is, the two are interconnected. A change in one results in changes in the other. You cannot independently change one without affecting the other without linear-phase filters/Xovers, such as exist within the DEQX (and within that, the other domain is still affected, but another gizmo applies reversal phase effects to cancel out the changes in the other domain). Time, frequency and phase are all STRONGLY related.

So Scops' statement was basically stating that by using an EQ to change amplitude over frequency (in the frequency domain) ALSO results in changing amplitude over time (in the time domain), as well as phase over frequency. You may perfect damping in one instance by building the right box (through Q), but if you apply EQ, you are almost certainly altering that damping at a specific frequency.