View Full Version : Spectrum analisys as a mastering tool

2009-01-18, 07:02 PM
How much attention, if any, should be paid to to the spectrum charts available in most audio mastering software?

I have a few recordings I have been tinkering with and have started experimenting with a technique that has been helpful to me and has helped me vastly improve some old cassette transfers and some more recent digital master files from my Zoom H2.

First, I have been using the spectragraph to determine where my recordings are underexposed and where they are overexposed (to use photography terms). I then use that information to slowly sculpt the sound, by chart only, to a near flat graph. Then I let my ears do the work from there. It has been successful for me on my own recordings and I was wondering if anyone else does this, if I am a dimwitted newb who should stop trying, or if this is a standard practice that I have stumbled upon.

2009-01-18, 07:18 PM
Nothing wrong with that. The home stereo manufacturers have been doing it for years using pink noise and a little mic to help customers eq the setup for their home listening space.

You find the deficiency and boost those freqs until it is even..

the big difference is that the pink noise gen as a source is 100% consistent on all frequencies, but your music source will not be... the amplitude of the frequencies will vary because music is inherently inconsistent.

In other words, how much you choose to even the amplitude across the spectrum will be subjective... like anything else with music / sound re-production.

I'd say pick some areas that look unnaturally dark and start there... but, a couple things to keep in mind:

The lower frequencies will always show more energy (amplitude), as it takes more to produce them. You should expect to see a natural fade from 0Hz to 20,000Hz, in amplitude - with the real drops coming at about 5-6k, then 10k, then again around 15k... don't try to get the same amplitude at 8khz as you have at 75Hz.. that will make it sound unnatural.

And as always... less is more. Don't do any more than you must.. your ears fatigue after just a few minutes of screwing with it. Best to listen, note BAD places, make an adjustment and STOP.. until later.. then repeat.

Less is more...
Less is more...

2009-01-18, 07:35 PM
Great info for me....thanks. I should have been a little more detailed in my process description.

I never boost with the onboard eq, I only attenuate the heavy areas. After I get the "sound" I want I then start the amplification and normalization processes (and what little compression I intend to use).

I have had problems with spikes causing my overall level to be much lower than I want. I do not want to use compression any more than 1 pass at about 1.5:1 or 2:1 at the max. I am afraid of "click removal" plugins, but it seems way too slow to goto each spike (each one being less than a few 100ths of a second long....but there are dozens of them) and manually attenuating to match the level of the surrounding audio. Is there an automatic process for this that I am missing in Audacity? I also have access to Pro-tools but hate to bother my friend who will let me use it free.

2009-01-18, 07:47 PM
Also, where can I find a screen shot of what is a typical spectragraph for a properly eq'd audience recording? I know each show will be different, just want to know how to know I am in the ball park.

2009-01-18, 09:33 PM
compare SA and FA with official releases... you'll notice typical amateur recordings have a hell of a lot of content in the highs that you would never, ever find on an official release...

and yeah, stay away from 'fix all clicks/crackles' plugins... if you want to fix them, do so one at a time, and don't delete lest you ruin the timing ;)

2009-01-18, 11:59 PM
Thanks Five....

never thought of ripping an official release and comparing SA.....thanks....

also, years ago I realized crack fix plugins/software sucked (at least the free/cheap shit does.....have a couple of old records i would love to put the waves plugins to)

2009-01-19, 01:04 AM
best way to fix pops is to highlight individual pops and run tegh click-pop eliminator on just taht one... one at a time. The most musical way to "correct" them. For non-pop (noise) peaks, you can just highlight them and 'normalize' to a bit above where the avg level runs. You are now screwing with the dynamics of the music, so don't drop it too much. It still needs to be a peak to keep the musicality of the recording.

What you should NOT do, that most of the boot companies do (and "pros" these days) is 'limit' the audio, then noramlize once the peaks are 'limited' off... nasty nasty nasty.

You gotta do them by hand to make sure you don't make the sound unnatural. Or... you could leave them and normalize to 99%... if they occured naturally in the recording, they are part of the sound... alter too much and the dynamics suffer greatly..

everyone has a volume knob... don't over do it!

2009-01-19, 11:08 AM
waves click/crackle remover sucks, too imo. best tool is adobe audition click/pop eliminator (fill single click now only not auto)... scroll down this thread for an old tutorial I wrote:

2009-01-19, 11:26 AM
What you should NOT do, that most of the boot companies do (and "pros" these days) is 'limit' the audio, then noramlize once the peaks are 'limited' off... nasty nasty nasty.

I respectfully disagree. It depends on what you're dealing with & how you do it.

I recorded a show in which the kick drum was ridiculous. I felt like I was being slapped in the chest 50' away. Limited the recording to just above the main bulk of the waveform, so the only part that was actually limited was the kick. Then normalized.

The problem with the boot companies & many others is they set the limit so ALL of it is limited. Might as well have just compressed the hell out of it, same result.