Dixie Chicks vs. Dream Theater vs. Floyd

I have long used reference EQ graphs from favorite commercial recordings as guides for my own mixes.

The one I've used longest is taken from Pink Floyd's Dark Side of the Moon, because it's one of my all-time favorite records and is musically compatible with my own style and instrumentation. However, DSoM seems darker than most modern recordings, and I began to wonder if I was being influenced by 1970's analog tape limitations.

There are a few websites that list recommended reference CDs, but very few of the recordings listed were in my own collection. So picked a title that was consistently included on these lists, Wide Open Spaces by the Dixie Chicks.

This is not a genre that I usually listen to and certainly far removed from any of my other references, such as Dream Theater or DSoM. But I was curious to see how it would compare when analyzed. I suspected that a well-EQ'd country recording would not be radically different from a good classic rock recording, but I was surprised at how close they were.

Here's a screenshot from Ozone's parametric equalizer, with two references displayed. The purple one is Dream Theater, the blue one is Dixie Chicks. The white line is a generic -6db/octave reference.



They are quite similar, except for one surprise: the Dixie Chicks have more bottom than Dream Theater.

Then I loaded up the Chicks in Har-Bal with the Pink Floyd reference, as shown below. The bright green line is Dixie Chicks, the lighter green line is Floyd.



Overall, there is only about a 1/4 db difference between the two! Except, again, the extended low end. (The peak at 800Hz is the lead vocal)

Curiously, the Dixie Chicks recording sounds subjectively brighter, even though the high end rolloff is nearly identical to the Floyd reference. Of course, there is a difference in instrumentation - lots of acoustic instruments on one, highly effected guitars on the other - and in the lead vocals, female versus male. But spectrally-speaking, the high end is very similar between the two.

Next, I compared amplitude analysis using Adobe Audition. In my own mixes, I shoot for an average RMS of about -11db to -12db. I've found that it's when I hit that mark that I get the best compromise between loudness, clarity and dynamic subtlety.

The Dixe Chicks hit -11.5db average RMS. Dream Theater, with their broad range of styles, aren't as consistent from song to song but tend to be quite a bit hotter, around -8db average RMS. Pink Floyd comes in about -12db.

While in Audition, I compared the Floyd and Chicks spectra. Again, there is a remarkable similarity between the two. In both cases, there is a slight prominence around 1.5KHz.

One of the measurements I look at is the db difference between the highest bass peak and at 10KHz. Above 10KHz there is a lot of variation in commercial recordings, but between about 100Hz and 10KHz is where they are most consistent.

Here are the db differences from 100Hz to 10KHz:
Floyd: 28db
Dream: 31db
Chicks: 27db

The larger difference between Dream Theater and the others is partially because theirs is a bit smoother on the bottom end, without the typical bump around 100Hz.

But Pink Floyd and the Dixie Chicks are nearly identical.

Now, I don't want to read too much into any of this. It's just interesting how much technical commonality there is between well-mastered recordings from different genres and eras.

Very interesting stuff!

I wonder what other people's favorite songs/recordings are for comparative analysis. I recently used John Mayer's "Waiting on the World" as a reference for a mix I was working on using Har-Bal and found it be very helpful. The songs were similar in style.

i think the dixie chicks album is misleading to listen to critically because the vocals are on the edge of harshness all the time. dark side vocals are so loose, deep, and dark sounding in comparison.

i think the dixie chicks album is misleading to listen to critically because the vocals are on the edge of harshness all the time.

I agree with this based on their "Taking The Long Way Home" CD, but I do not have the one bitflipper used in his comparison. "Long Way" hurts my ears to a degree due to the harshness of the high freqs - or maybe like you, it's my age!

the Dixie Chicks have more bottom than Dream Theater.

I think it's the little chubby lead singer that throws them over the edge.

You do have to be careful to analyze a representative part of the song.

True. I forgot to mention that when I take a spectrum for a reference, I chop of the intro and fadeouts and average the main body of the song. Or, if it has an extended solo section like all-acoustic guitar or a long drum break, I leave those parts out of the analysis.

This is especially true for some of my favorite bands, who often start a song in one style and end it in another. Dream Theater does that a lot, with thick synth pads at the front, morphing into straight-ahead metal later on. Such songs do not make good references.

...I think The underlying effort was to go for smooth, although the music sounds great, to me the mix needs a freshen someday...

In the DVD documentary about the making of the album, Roger Waters said that the the final mix (done by Chris Thomas) was much warmer or mellower sounding (I can't remember the exact words he used) than they had intended.

Can someone explain how to use one of these graphs and exactly what your guys are talking about?

I think I get it. Am I correct in saying you a looking at a graph of what frequencies are present and at what level at any given point in a song? Then comparing that graph to a slice or specific point on your own project? Or is it an average of the frequencies and their levels based on the whole song?

And where do you plug in this software to use it? Does it go in an effects bin on your master bus, or somewhere else.

Thanks

In the DVD documentary about the making of the album, Roger Waters said that the the final mix (done by Chris Thomas) was much warmer or mellower sounding (I can't remember the exact words he used) than they had intended.

Softer and homogenous were the terms he used. And of course Alan Parsons also says that Thomas used more compression than he would have done.

There's quite the upside down bell in the low frequency (40-60hz) range on the first graph. How do the recordings feature so much low end there and avoid being overly boomy?

It's not too uncommon to see that, though often it's pushed up further than that. It's not uncommon to see a lot of emphasis on the second and third octave with the first rolled off heavily. And sometimes boomy doesn't have anything to do with the amount of bass, but the amount of bass combined with the other instruments and how they interact with the bass, how the bass is played, and how compressed the bass is. If the bass guitar is the only thing down there, and everything else is high passed up pretty high, not uncommon, there can be quite substantial bass and it won't necessarily sound muddy or boomy.

ORIGINAL: bitflipper

I used two different products for the above screenshots. The first is Ozone, a plugin that's a collection of tools, usually used for mastering (so it typically goes in the effects bin on the master bus). One of its components is a parametric equalizer with a spectrum-matching feature. I don't actually use it for spectrum matching, but I always display the -6db/octave reference and often another reference from a commercial CD.

I do the same and it has helped me alot. I definitely feel that my mixes compare more favorably now. I'll look at some more when I get home this evening, but I know that Michael McDonald's remake of "Aint No Mountain High Enough" does not roll off much at all on the highs; it goes out pretty straight and does not slope much at all. I'm going to put some others on it this evening.

Using the spectrum analyzer really helped me alot though cause I would always wondered why my mixes were so dark, then the analyzer revealed that things were dropping off way too much in the highs. It's helped me set levels better and provided another means of feedback.

From a practical standpoint - after comparing a recording of yours to one of these as a "guide", where do you go from there? If your recording was sounding good to you before this comparison, and then you see that it's pretty similar to one of these reference recordings, I imagine you are basically using it as a visual confirmation that your ears are hearing it correctly and would probably keep your mixing/mastering, correct?

If there's a bigger difference, what do you do then? Do you try to more closely match it? Do that just signal to you that it warrants a closer listen to those frequencies where there are differences?

Also, you mentioned "One of the measurements I look at is the db difference between the highest bass peak and at 10KHz." Can you please explain why this is important?

From a practical standpoint, it is rarely advantageous to try and duplicate someone else's spectral balance exactly. Even two songs with the same instrumentation are going to vary, and most of the time, EQing to some arbitrary standard just isn't going to work very well. The idea is to identify where you differ from the norm, and then determine by experimentation if that divergence is good or bad.

However, even if they're of limited value as literal templates, references do serve several practical purposes.

First, if you don't have an ideal listening environment (and who does?) it'll give you a visual check that may indicate when you're compensating for your room rather than your recording.

Second, it shows you the normal range of what's acceptable for different styles of commercial music. If you see that your curve is steeper in the high end dropoff than commercial references, then it's likely you're rolling off too much high end - perhaps because your monitors are overly bright. Conversely, if you see that your low end is hotter than most professionally-mastered CDs, maybe it's because you need a subwoofer to augment the low end of your monitors.

Although I don't advocate trying to precisely match EQs, that doesn't mean you can't try adjusting your EQ in the same general direction and auditioning the difference. Sometimes you'll be surprised to find that a half-db dip around 1.5KHz makes the whole thing sound better. You might not have thought to try that until you compared your mix against some high-quality recordings.

[I should insert another observation here: always try to fix EQ in the mix first, and only then resort to equalizing the master bus. EQ is all about balance, and that means the balance of instruments, their relative tones and their relative levels. It's actually possible to get a nice EQ without even having an equalizer across the mains! Not easy, but entirely possible.]

As for measuring specifically between 100Hz and 10KHz, it's because that's where most of the meat is, where the spectrum is most linear, and where commercial recordings tend to be most consistent.

What I'm looking at is the slope of the curve as it slants downward left to right. That slope largely determines the overall brightness or darkness of the song. A shallow slope will give you a bright-sounding pop mix; a steeper slope will be mellower.

There are two generic guideline slopes: -6db per octave and -3db per octave. The former is a more classical slope, and is actually based on the natural EQ slope of a classical orchestra. The latter is more for pop music that'll be played on cheap iPods and AM radios, accentuating the vocal above everything else. Your own music may work better with one or the other, or something in between.

That's not to say that the 100-10k region is the whole picture. Below 100 and over 10KHz are just as important. It's just that for comparison purposes, there is too much variance among commercial recordings at the extremes to draw much practical guidance from those regions.

It's one seriously fancy EQ. And a great learning tool.

Yes, I just bought Har-Bal recently and it is very cool. However, I wish it was a plug-in that could be placed on the Master Bus (as you do with Ozone).

I'm sure you're familiar with Voxengo SPAN. Is it possible somehow to plot an EQ curve in SPAN to be used as a reference?

Unfortunately, SPAN lacks the ability to save and restore snapshots.

Too bad. It would be really nice to have an "ideal" EQ curve (if there is such a thing) plotted in there to use as a reference throughout the mixing and mastering stages of a project.

Yeh, probably so. Most commercial stuff these days is compressed to death, so it won't have much dynamics. It's best to compare to soemething that also has the same type of dyanamics as you are shooting for, as well as the same spectral balance.

I guess one of the great joys of doing it for fun is that you can create music that has a lot of dynamics and if people don't like it, screw'em, they aren't paying you for it anyway. For that matter, they aren't paying the people making the hyper-compressed music all that much either.

one of the great joys of doing it for fun is that you can create music that has a lot of dynamics and if people don't like it, screw'em, they aren't paying you for it anyway

My thoughts exactly, Dean. In fact, I'd rephrase your comment as "you can create music that has ___________ and if people don't like it, screw'em, they aren't paying you for it anyway." Fill in the blank any way you want: tubas, monkeys, political content, weird time signatures or nonsense lyrics.

Making music for fun is liberating. It surely beats singing for your mortgage payment, which I have done.