EQ and Compression for Podcasters (Who Have Never Used Them Before)
I wrote an article about the simplest workflow in achieving a studio quality mix and edit for your podcast. In it, I listed EQ and compression as the most important audio effects you need to learn. But I realized that a discussion on those two would require a wordcount that is equal to a moderate-length article so I decided to write this separately.
This guide is applicable to all brands of audio editing software but I would highly suggest you look for a DAW (Digital Audio Workstation) software where you can apply the effects non-destructively and hear the resulting sound as you adjust the parameters of the EQ and compression (real-time preview). Not only does it make things easier, it also makes you understand the effect much faster as you start to correlate each adjustment with what you are hearing. For Apple users, Garageband fits, and it’s free. For PC users, I understand that Audacity has become sort of a ‘go-to’ for editing podcasts, but it’s EQ and compression plugins isn’t non-destructive and lacks real-time preview. Installing a third-party EQ and compression does enable live preview but it doesn’t run very smooth and still destructive. A quick google for free DAW will yield some good ones. In my case, I have been a Reaper user forever. They actually offer a non-crippled trial of the software. And if you decide it suits you, it comes with a very pretty price tag, USD 60 for non commercial use (I am not in any way associated with the Cockos, the maker of Reaper).
SO LET’S START WITH EQ
A human ear can hear tones from the very low 20 hertz up to a very high tone of 20,000 hertz. An EQ is basically a graph of that, it starts from the left side at 20 hertz running towards the right which is 20,000 hertz. So low tones in the left, mid tones in the center, high tones on the right. In the middle of the graph is a horizontal line you can manipulate and add all sorts of upward or downward curves. Using an EQ is basically modifying that line to either boost or cut a certain frequency. Say your recording has too much bass on it, draw a downward curve on the line at around 100 hertz on the left, or if you need some more midtones on your voice, drag the line to upward to make a curve at around 2,000 hertz.
So where do I need to cut and boost?
If you’ve done enough prep work on our recording, you probably won’t need to EQ at all. But say you’ve done well on your own recording but your guest audio wasn’t recorded as well as yours. Or after a week you realized your voice sounds boomy on the recording. Or maybe you checked your reference podcast and you feel like your sound isn’t as close to that. Which frequencies should you boost and which frequencies should you cut. You can probably google some rough guides on EQ-ing like a table that says if it sounds muffled cut at 400 hertz, or if you need some ‘sparkle’ boost at 10,000 hertz, but we won’t do that here. Instead we are going to learn how to sweep. The simplest way to sweep is to make an upward and narrow curve on the EQ to boost a very narrow set of frequencies, then while playing the audio, slowly move that curve from left to right (from 20 hertz to 20,000 hertz) until you nail which frequencies you want to cut or boost. (see image below)
If your muddiness gets worse at around 300 hertz, then you now know the culprit. Once you nailed your target ‘muddy’ frequency, you can just drag your upward curve down to make a cut.
So you see, EQ is basically volume knob, but instead of raising or lowering the volume for the whole audio, it only raises the volume for certain tones in the 20 to 20,000 hertz audio spectrum.
If EQ is a precision volume knob for certain frequencies, then compression is like a robot that turns down the volume knob for the whole audio (not certain frequencies).
Have you ever had that time when you are watching a movie at night and you’re afraid you’ll wake your roommate in the next room, so you make sure to turn down that volume whenever a loud car crash or explosion happens in your movie (you love watching Michael Bay movies, what can you do?). A compression plugin does that for you automatically. Like a robot with its hands on the volume knob, it will lower the volume when it gets a little too loud. But unlike a human, it doesn’t know what is considered loud and how much it needs to dial down so you need tell him very specific instructions like how much loudness is acceptable and by how much should he reduce any loud sounds.
For that, we need to tell Compression four basic instructions: Threshold, Ratio, Attack, Release.
Threshold is like telling your robot to start turning down the volume once it reaches this loudness. So for example if your threshold is set at -18 dB but your audio climbed up to -16 dB (remember: as the number gets closer to zero, it gets louder), then compressor will start turning down the volume.
So what should be the threshold for your audio? Easiest way is to look at the waveform of your recorded audio, find the peaks where you are speaking naturally (not the screaming and laughing part), then set the threshold just a tiny bit below that. That way, when you are speaking with your natural voice it gets compressed a tiny bit but when the laughing and screaming starts and jumps over that threshold, all that sound gets compressed heavy. What you get is a very smooth levels but still has enough dynamics in it.
Ratio is like telling your robot how much should he dial down the knob when the audio is going past the threshold. A quarter turn of the volume knob? Half turn? Let us say you gave the compressor plugin a ratio of 2:1. An easy way to understand this is to switch the two numbers and turn it into a fraction, so that 2:1 now becomes ½. So for example you have a section of your audio that peaks at -16 dB and you have your threshold at -20 dB. That means you have 4 decibels of audio going past the threshold. If you have a ratio of 2:1, flip the numbers and you get ½. One-half of 4 decibels is equals 2. That means the original 4 decibels above the threshold will be compressed to just 2 decibels. So your -16 dB peak is now reduced to -18 dB.
Attack is like telling your robot how fast you want him to turn the volume knob. Should he kill the loudness really quick (fast attack) or should he dial it down gradually and smooth (slow attack). Attack is described in milliseconds.
Release is like telling your robot how fast should he return the volume knob to its original position once the loud explosions in the movie is over. Same with attack, Release is described in milliseconds.
So what’s the most common compression settings to use?
This is really hard to answer because every audio I have worked on required different compression settings. But I do have usual settings that I go to to start out specially when it comes to a recorded speaking voice (in a podcast).
Whenever I encounter a very controlled speaking voice, someone who speaks on a constant volume (you can see it in the waveform too, an example below), I usually set the Ratio at 4:1, with a 3ms Attack and a 100ms Release, and then drag the threshold slightly below the lower peaks until it starts to compress regularly some 6 dB of sound.
Whenever I encounter a very dynamic voice, someone who whispers the first seconds then go loud the next (the waveform is uneven and with lots of very high peaks), I usually set the Ratio between 5:1 to 10:1, with a 3ms Attack and 100ms Release, and then drag the threshold near the lower peaks until it compresses some 3 to 6 dB of sound.
One last note on compression
Since compression is basically dialing those loud peaks down, you may notice that the volume of a compressed audio goes down significantly (unless there is an auto make-up gain button on your compressor, in that case you should turn it off). Your original peaks of -16 dB will usually end up lower as most of those peaks get compressed. So it’s very important that you lift the output gain of your compressor so that we go back to the original peaks of -16 dB.
A discussion on compressor isn’t complete without the topic of limiters. The main application of limiting in editing podcasts is to lift the overall volume of your episode without going past the 0 decibel mark (distortion). Meaning, if you simply add more gain, or drag the master fader in your editor, there’s nothing that would stop those loud peaks from hitting the 0 decibel ceiling leaving you with a clipped sound. That’s what limiters are for.
Limiter is just the same as a compressor but with really high ratio (10:1 to infinity:1), fast attack (0 ms), and fast release (70 ms to 200 ms), so that a very minimum to almost none of the peaks will get through the threshold. You then set the threshold just below the high peaks and then set the output gain to hover just below the 0 dB mark.
Fortunately, there are standalone limiter plugin with simpler controls so that you won’t have to deal with all the compressor settings to turn it into a limiter. You simply tell the limiter plugin how loud should the maximum peaks be (ceiling) and which peaks should be limited (threshold).
So that’s it, you now have an introduction on the two of the most important processing plugin in mixing. Your next step is to record a few minutes of vocals and start trying out some of the techniques and setting I wrote here and most importantly, LISTEN to what it does to your sound so your brain can easily relate what a certain knob does, making it easier for you to learn this new concepts.
Argh! This will probably sound too technical for some even though I tried to explain it in the simplest way possible. If there is something I didn’t explain very well and only made you more confused, feel free to ask in the comments.