-6dbl slope (and what RMS is...)

I just don't get it
I've analised commercial cd's, I've analised the -6 slope file steveg posted and I've read the ozone manual file that was posted and I just can't see the slope or I do but I don't know it.
on the -6 dbl file steve posted I see a difference  from -34 to -84 dbl from one end to the other am I supposed to be looking across the entire frequency range or is there an area that this slope refers to
I know there are many threads on this and I have read and read but I'm just not geting it. if possible could someone explain what I am looking for based on the -6dbl file steve g posted (I believe it's on zemlins site) and then how it applies to other wave files
I appolagise to the ones that are sick of trying to explain but try to remember the times when you had trouble understanding things

thanks

I appolagise to the ones that are sick of trying to explain but try to remember the times when you had trouble understanding things

I understand this perfectly - probably rather more than you might think. So here are some different words - see if any of these help at all:

The slope starts from about 1kHz, and then slopes down at 6dB/oct(ave) - so as the frequency doubles, the output halves - this has the effect of leaving the amount of energy in the upper part of the spectrum the same all the way across - the idea is that if you have a signal at -5dB at, say 2kHz, then the same strength signal at 4kHz would have to be at -11dB to be percieved as being 'balanced' with the lower frequency.

So if you apply that idea to a mix, then the general 'slope' of the top of your mix, when viewed as a spectrum, would be downwards at 6dB/oct. Now, a lot of people don't think that for modern music, this slope sounds bright enough, and they've proposed that a 'pink' slope is more appropriate. This slopes down more gently, at 3dB/oct and leaves the higher frequencies more pronounced than they would have been with the 6dB/oct slope.

You can't 'use' the slope as it is - you have to view your mixes against where it would be - and because signals come and go somewhat, this isn't always the easiest thing to do. The idea is that as long as the overall sound sort-of looks as though it's sitting on the slope line most of the time, you won't be far out. The way this is often achieved is with broad sweeps of EQ - nothing too peaky, otherwise you will hear the effect in specific places, rather than having the whole mix rebalanced.

It is also very possible that programs like HarBal will help you to achieve this balance more easily, although I can manage it perfectly well with Ozone, which has some helpful bits - like 1/3 octave displays with a longish peak hold. This gives you a much better overall picture than you would get from just Audition's spectral display alone.

I 've got it thanks steve
so if it starts at 1khz and works up the frequencies it affect the highs the slope gives a nice streamline volume so the higher up frequencies don't pierce your ears (i think)
what about below 1khz? is there a standard or is it all ear there because its eaiser to hear the rumble and thumps

ob - Auditions spectrum view probably moves too fast for you to get the slope in your minds eye. SteveG mentioned he sets peak hold longer in Ozone which lets the display hold the pattern a bit longer so you can see it. That lets him adjust his curve to the max or peak of the program material envelope. You can also set your average time longer so that the general slope or curve becomes more apparant as it 'averages out' a bit and will hold still long enough for you to get a good glimpse of it.

I don't know if you have Adobe Audition 1.5 but here's a free VST Spectrum analyzer plugin that will let you look at things like that at full resolution or 1/6 octave or whatever while displaying rms values and whatnot (your next step is rms I'm guessing in your reference checking of commercial CDs):

http://voxengo.com/freevst/

It's called SPAN and is about the 6th one down...

I forgot to mention that SPAN has a slope you can set from 0db/octave upto about 8dB/octave I think - the pivot point is at 20Hz unlike Ozones 1KHz reference pivot.

Pink noise has a 3dB/octave slope just to give you a reference. A lot of rock music I listen to also has a 3dB/octave slope (from 20Hz). So in other words if you set the SPAN spectrum to have a 3dB/octave slope and set the display at a reasonable speed then you will mostly see a flat  envelope along the horizontal axis on a lot of rock material.

I don't know what orchestral stuff looks like (or how to spell it, hehe) - you'll have to ask SteveG, Oz, and Zemlin - folks like that !

I 've got it thanks steve
so if it starts at 1khz and works up the frequencies it affect the highs the slope gives a nice streamline volume so the higher up frequencies don't pierce your ears (i think)
what about below 1khz? is there a standard or is it all ear there because its eaiser to hear the rumble and thumps

In theory you are right - it should extend down as well - but for purely aesthetic reasons it doesn't. You have to bear in mind that the 3 and 6dB curves are only a composite view of what has been generally found to be the average responses of a large number of tracks of different types that have been judged to represent the sort of balance that people like to hear. I think that there are a number of factors involved for why the response is flat up to 1KHz - partly rooms and loudspeakers, partly our ears and the Fletcher-Munsen curves, resulting in the simple fact that if you extend the curve down at the same rate but from a lower frequency and mix to this, you get a very bass-heavy and wooly sound!

I forgot to mention that SPAN has a slope you can set from 0db/octave upto about 8dB/octave I think - the pivot point is at 20Hz unlike Ozones 1KHz reference pivot.

Apart from wondering just how much 'slope' you get at 0dB/oct   I don't think that having it extend down from 1KHz is any help at all for this purpose, for reasons just given...

Pretty minute slope setting at 0dB/Octave eh ?  

so there is light at the end of the tunnel I've been looking at some of my original stuff from over the years and its amazing how expierence pulls youi closer to the slope.
I guess your right kylen  RMS would be my next question
you guys are doing a grteat job its making sence to me
once we get RMS out of the way may be someone could explain how to curve the peak I get from the girlfriend for spending to much time in the studio?

RMS would be my next question

What about it?

what is it, how can I optomize it

[R]

what is it, how can I optomize it

Don't really understand the question - well not the 'optimise' bit, anyway... RMS stands for Root of the Mean (value) Squared. It's a way of always giving a positive value to a waveform. It's useful because it indicates the amount of power (area under the curve, if you like) rather than the instantaneous peak value of a waveform. To do this, the values have to be measured (and calculated) over a finite amount of time. With a steady sine wave, the value is relatively easy to compute - you'd take the mean value of the sine wave (which would technically be the average value of several equidistant mid-ordinate points in one half cycle, but don't worry about that), square it, and then take square root of the result. Because you are squaring a number, it doesn't matter whether its a positive or a negative half-cycle you calculate with - you'll always get a positive answer.

The history of this is all to do with power, and how to calculate a value for an AC power that would indicate the same heating effect that the same value of DC power had. Turned out to be quite simple with sine waves, but when you start to do it with non-sine waves, it gets a little more tricky. The explanation about the power side of this is here (scroll down the page). Suffice it to say that when Audition does RMS calculations, it has to use an entirely different stategy that takes account of all of the arbitrary waveforms that it finds.

What makes RMS useful for audio is that it relates very much to the way people percieve volume changes - which is why it's the preferred way of running the detector in the Dynamics Processor. But because of the way it has to take account of the mean value of a signal, it sometimes appears to ignore the odd spike - and I can only assume that this is entirely as a result of doing the sums correctly!

There's nothing for you to optimise, except your understanding, I think you'll find...  

that made sence thanks steve