Human hearing of frequencies is calculated on a scale of hertz or cycles per second.  The commonly defined range of human hearing is 20 hertz to 20,000 hertz per second.  A graphical EQ device, be it a hardware version or a software plug-in like those shown here generally work within that range. (Depending on your age and life experiences your hearing may be markedly less than this…)   In the application of boosting or cutting particular frequencies there are also a few controls which determine how the boost or cut will be applied.  The first is the particular frequency itself, the second is the amount of boost or cut, usually described in decibels, and third is the “Q” or width of the boost or cut.  If the horizontal line is flat, as shown here, then no EQ processing is happening.

10 band parametric EQ

Looking at the photo here you can see the controls just described.  This particular EQ is a 10 band EQ meaning that you can affect 10 different bands or frequencies at one time.

The graph indicates the frequency to be affected sorted with the low frequencies at the left of the graph to the high frequencies at the right.  Each ‘band’ has several controls – on/off, the type of “curve” to be applied, the gain or amount of the processing, the frequency where things are being applied, the Q of the processing and the overall input of the source and the output after processing.

Bell EQ curve

Here is a bell curve which looks sort of like a – bell.  This is a very common way to apply EQ to a particular frequency.  The Q is set to very wide so it is affecting a wide range of frequencies that are near the key frequency.  Applying EQ to satisfaction is to determine the central frequency to be affected, the size of the Q and the shape of the curve and the amount of gain that is being cut or boosted.

Shelf EQ curve

Here is a shelf curve that sort of looks like – yes – a shelf.  Different EQ makers apply this differently and this one makes a slight cut prior to the boost.  By the way these are pretty extreme boosts being made in these examples to more clearly show what is being discussed.  Often EQ will be applied at much lower boosts or cuts and at several different frequencies.  The idea with EQ is generally to do as little as possible to affect the desired change.  You can easily mangle the recorded audio to something unacceptable.

Low Pass and High Pass filters

This picture shows a different type of EQ – the curve at the left is called a “High pass” filter and the curve at the right is called a “Low pass” filter.  These filters will remove all of the frequencies at the cut off frequency.  So, in this example, the high pass filter removes all frequencies below 100 hertz and the low pass filter removes all frequencies above 10,000 hertz.  These are very useful filters for removing unwanted elements of recordings.

These “pass filters” are powerful tools for affecting the recorded audio being processed.  Here are a few examples that might shed light on their particular usage.  Let’s say that your boom man has trouble holding the pole and his fingers are creating small “boom” sounds during the recording.  Often these noises are very low in the frequency range and are difficult to hear without a sub-woofer.  A high pass filter may remove all of the noises without affecting the quality of the recording.  Or perhaps the tracks sound a bit “murky” so a slight presence boost at 2.5K or 3.5K may help “lift” the voice and a small cut at 300 – 400 may remove a bit of “boominess.”   If you look at the EQ that has a low and high pass filter applied you’ll see that it creates a “band” of frequencies that will pass through it.  The other name for this sort of setting is a “band-pass” EQ and these can useful in a number of ways.  Since band passes limit the frequency range of a recorded signal they are useful for many things. Many devices such a telephones have a “limited band-width” so you can mimic this bandwidth with a band pass filter. 

Notch at 1K

Let’s say you hear a steady noise in your recording that is present in the upper frequencies.  You may want to try a “notch filter” which can notch out the offending frequency.  Here several bands are tied together to really define the notch.  The idea here is to carve out only the offending frequency and try not to disturb anything else around it.  As you can see a graphical EQ is a very handy way of visualizing your sound.  Mix engineers will often “sweep” the notch along that horizontal range of frequencies until the frequency range to reduce or add is pinpointed.  Notches can be very useful in eliminating any type of steady state noise.  Steady is the key idea because if the noise or sound oscillates to other frequencies then the notch is no longer relevant since it is specific to only a narrow range of frequency.

EQ is an amazing tool with many useful applications.  However it is not a miracle device and it is most certainly not a noise reduction device.  It can indeed reduce noise but it is less successful at that than a dedicated noise reduction device is.  There are “broad band” noises such as air conditioners that cross many frequencies at one time.  They usually can not be notched or pass filtered out.  They will require some different processing, but if you need simple boosts or cuts or if you have tones that need to be minimized or emphasized in your recordings that are steady state, EQ will be the right tool to use.

This is merely a quick explanation of this wonderful, misunderstood and useful tool.  Experiment with EQ and learn how each of the controls affect the frequencies.  Learn how cutting and boosting achieves different effects and how specific frequencies affect specific sources.  Listen to the world at large and imagine if you had to use EQ to recreate the sounds you hear.  Walk by a stone building with music playing inside, loud but boomy and distant, what EQ curves might be applied to a standard music track to achieve a similar sound?  Listen to the world around you. This in time will guide you on how to use EQ to mimic the sounds in and of everyday life.  And then use those ideas to create the sounds you hear from this world – or others. Comment with thoughts, ideas or questions at will.

I was recently mixing a broadcast project with some serious audio issues.  These production issues are always a head scratcher.  There was an interview of two subjects sitting side by side.  One was distorted and sounded like they were in a wind tunnel, the other sounded clean.  The scene cut between a two-shot and individual close-up shots of the talent.  None of the audio cuts even remotely matched even though they were sitting next to each other!   Blasts of clean audio cut to distorted audio cut to the wind tunnel.  I was asked to “clean it up a bit.”   Easier said than done!

I started checking through my box of tools to tackle the problem and several fit the bill.  I won’t name the other programs but one of them wouldn’t authorize although I’ve owned it for over a year.  Their website was useless for this and they wanted me to pay for support.  I can’t quite figure out how these companies have the gall to charge me to figure out why the program I bought and paid for won’t load.   But … that is a rant for another day!  Let’s just say that I have had these problems with this company before and now I’ve finally “washed” my hands of their noise reduction product and them.  (hint, hint…)

RX has five separate modes for tackling tough audio problems.  These are not unique to iZotope since these are the types of tools generally found in similar sorts of programs.  However the iZotope implementation has a wide variety of parameters to adjust how the source file is being effected.  These are very powerful tools with a real range of abilities.  A brief description of each is -

Declipper – a repair algorithm that finds and repairs analog or digital clipping or overmodulation.

Declicker – a repair algorithm that finds and repairs clicks and crackles from recordings.

Hum removal – a repair algorithm that finds and repairs hum and buzz.

Denoiser – a repair algorithm that finds and repairs pesky broadband noise problems.

Spectral Repair – a repair algorithm that finds and repairs random noise within an audio file.  Pretty cool feature.

Unfortunately for me I had to use each of these tools on this interview.  And they rocked it.  The hum removal is an adjustable comb filter that takes out the main frequency hum and it’s associated harmonics.  (Comb filters are filters of many small bands that graphically look like a – comb.)  This has many parameters to adjust and worked like a charm right off the bat.  Next I had several bits of horribly distorted audio.  Using the declipper tool I was able to make things sound less awful.  Not great but definitely better.  I can’t blame RX on this one, fingers pointed squarely at the sound recordist.  Each of these tools have a box where you can audition only the portion of the audio which is being filtered, adjusted, manipulated or repaired.  I wanted to check that the decipper was actually addressing the issue and when I listed to only the audio to be repaired lo and behold all I heard was distortion.  So although it didn’t “fix” the file it took some of the edge off.

Next up was the A/C – wind tunnel problem.  This was a real treat.  I used the denoiser tool to clean these audio bites up.  As is the case with other similar tools on the market you find a “clean” portion of the noise (huh??) and “train” the filter what to remove.  Once it’s trained you process the file.  There are a couple of denoiser algorithms and I found in this case the “offline” ones worked best.  I am working in ProTools and you can run the RX tools as inserts to process the track in real time.  Since these had so many different noise problems I hard-processed the files and rewrote them.  These higher powered algorithms are not available as a real time insert and instead process offline – or in other words rewrite the file with the processing.  The difference was stunning.  All of a sudden I was able to match the audio files to sound like they were actually – recorded well!

This is a highly effective tool for audio post.  I have not used it specifically on music sources but if you check their website out you’ll see several powerful demonstrations specific to music.  I have used all of the major software vendors for noise reduction as well as the high-end hardware units.  Now this can’t compete with the dedicated hardware solutions but it’s not thousands of dollars either.  If you find yourself up against difficult audio restoration in your projects I would highly recommend RX as a solution.  These tools can make you a hero to the producers who want a “little clean up” on their horrible audio recordings!