MeldaProduction Dynamic Equalizers: MDynamicEQ – Part 2
MDynamicEQ offers excellent metering, displaying several useful pieces of information in one panel. Meters may be hidden when not needed, displayed alongside the main window or in a floating window to save horizontal screen space.
Click on the little button next to the preset selector to show the meters alongside the main window. The button then moves down to the bottom of the meter area; click it again to hide the meters.
To float the meter window, click on the third button from the bottom in the lower-left.
In and Out are self-explanatory. They’re standard RMS/Peak Hold meters. The numbers at the bottom of each are the momentary peak levels. Click on them to reset.
“Side” and “Width” require a little explanation, as you’ve probably not seen such labels in other plugins, or not with the same meanings.
“Side” is an abbreviation for “External Side Chain”, and shows the sidechain key’s level. Note that this meter has no meaning when the external sidechain is not being used.
In other plugins, the Width meter might be called a Correlation Meter. It shows stereo width: the amount of difference between the left and right channels. The greater the difference, the more width you hear. When used on the master bus this can give an idea of how wide your mix will sound. It can also highlight phase-inversion problems.
When the Width meter reads at the bottom of the scale (0%), it means there are no differences between the left and right channels; in other words, mono. When the meter reads at the top of the scale (“Inv”), it means the left and right channels are exact mirror images of one another. That’s usually an accident, is usually undesirable and means your mix is not mono-compatible.
A nice, wide music mix will bounce around from about 30 to 70 percent, depending on the style and genre. The preferred range between 66 and 100% is colored blue, but don’t worry if your mix goes into the lower green area, as long as it isn’t stuck down there all the time. Also don’t sweat the occasional incursion into the red (> 100%) zone, as long as it doesn’t stay there long.
Note that by default, this meter uses EBU pre-filtering to simulate human perception, which helps get a realistic appraisal of how wide your mix sounds. However, as a clinical diagnostic it’s best to turn this off so as to give equal attention to all frequencies. Right-click on the meter to toggle loudness pre-filtering.
In addition to the global meters, there are also meters available for individual filter nodes that show instantaneous and peak levels, plus instantaneous and peak gain reduction/expansion. To access these meters, right-click on a filter node to bring up the filter settings dialog. Unfortunately, it’s not possible to view more than one filter at a time, but I don’t think this is a big deal because you typically only look at them when fine-tuning the dynamics.
OK, enough with the technical details. Let’s walk through some real-world usage scenarios.
MDynamicEQ is suitable as either a mixing or mastering tool, for both surgical corrections and broad-stroke coloration. It’s more CPU-intensive than most general-purpose parametric equalizers, so it may never become your go-to track EQ, but it has a surprisingly number of diverse applications for which it is uniquely suited. Sometimes, the extra CPU load is offset by eliminating the need for a compressor that otherwise would be inserted in front of a standard EQ.
For the first example I’m going to insert MDynamicEQ into a vocal track that has some issues with room resonances, the result of having been recorded in a too-small, too-live, too-square room. This is may be the best example of an application that specifically suits a dynamic equalizer better than any other conventional approach.
Using MDynamicEQ #1: Mitigating Resonances
When you first insert the plugin, you’re initially looking at an edit window with 5 disabled nodes. (If you don’t see them, click the Edit button at the top to enter the main editing view.) Double-click on one of these five nodes to enable it. Move it up and down to set its static gain, move it left and right to change its center frequency, just like any garden-variety parametric equalizer.
Sidebar: Static Low- and High-Pass Filters
In addition to the five dynamic filters, there are also two static filters that aren’t immediately visible: 12 dB per octave low-pass and high-pass filters, whose slopes can be chosen from a fixed list of values from 6 dB/oct to 120 (!) dB/oct. The high-pass filter is especially useful, on both tracks and busses, for taking out subsonics or reducing low-frequency mud.
These are pretty mundane filters, so there isn’t much to say about them – except for one non-obvious tip: how to activate them. To do that, drag from the left or right border of the display.
There are little “arms” stretching out vertically and horizontally from the node. The vertical arms adjust the dynamic range and the horizontal arms adjust the width, or Q, of the filter. You can either drag them with your mouse or, for the Q adjustment, use the mouse wheel to widen or narrow the filter.
In this image, we see two filters with differing bandwidths. Note the vertical arms, which show that the left filter is static (arm is centered on the node), while the right filter is using downward gain reduction (arm descends below the node). Both parameters may also be set via the filter settings dialog if you need to enter precise values.
For this particular application, the first thing I want to do is identify the resonant frequencies that need to be tamed. For that, I’ll switch on the sonogram display and play back the track.
The sonogram display might look like just a pretty gimmick, but it’s actually a very helpful visual representation of the changing frequency content of the track. The vertical scrolling over time makes it easy to position a filter over some significant frequency, such as the bright green splotch in the image above.
The first sonogram feature you’ll want to discover is the Pause button, located in the upper-right of the edit window. This freezes the sonogram display. It’s very helpful when you’re trying to identify a short-term event such as a resonance that only occurs when certain notes are sung or played.
In the picture above, we can see that there is an occasional bright spot occurring at about 100 Hz. This is the resonance we want to address. All we have to do is listen for the resonance during playback, freeze the sonogram and place one of the filter nodes atop it. Drag the horizontal (bandwidth) arms so the filter is wide enough to cover the resonance but doesn’t encroach on adjacent frequencies.
And as easy as that, we’ve now got a filter in exactly the right spot to catch that resonance. No hunt-and-sweep, no guessing.
Now that we have the filter positioned, we can set up its dynamic behavior. In this case, our 100 Hz resonance is only a problem at certain points in the track, so we’re not going to add any static gain change like we might do with a normal equalizer. Instead, we’re going to have the filter kick in and apply dynamic gain reduction only when the level gets out of hand.
For that, we’ll adjust the dynamic range by dragging the vertical arm downward to enable compression. I’ve decided that it needs a maximum of 12 dB reduction, so I’ll drag the arm down until it reads “-12 dB”. This determines the maximum reduction that’ll be applied: 12 dB. Alternatively, if you want to enter a precise value, right-click on the node and enter “-12” in the Dynamics setting.
The actual amount of gain reduction will be determined by the amplitude of the frequency band at any given moment in time, and whether it’s above or below the threshold setting. By default, the threshold is silence, making the filter’s dynamics active all the time. Leaving the threshold at silence actually works fine in most situations.
Let’s assume an always-active filter isn’t what I want, in which case I’ll set the threshold to a little below the nominal level of the vocal track. That way, the filter will be largely transparent, applying gain reduction only when needed. So how did I determine where the threshold should be? I used the Analyzer display to observe where the band sits normally versus how high it jumps up when a resonant peak comes along.
At this point, let’s take a brief detour and mention the Analyzer before proceeding with the resonance mitigation example.
At first glance, the spectrum analyzer may seem very similar to others you’ve used before. But it’s got some nice features, including a few that aren’t at all typical, that make it a standout spectrum analyzer. Useful features include:
- The very helpful Pause button
- Optional 1/3 octave and 1 octave modes
- View Pre, Post or Sidechain
- Variable opacity
- Peak detection
- Frequency and note value of peaks
- Optional pre-filtering
- Auto-listen (sweepable bandpass to audition individual bands)
- Areas (color-coded divisions on the graph identifying frequency ranges)
Of all the features listed above, one of the coolest is the semi-transparent overlays that show the frequencies (and note values) of the most-significant peaks. This will allow us to get the precise frequency of the resonance we’re stalking.
In this example, we see that the peak is actually 86.5 Hz. The grid labels aren’t visible in the screenshot, so you’ll have to trust me that it peaks at -6 dB and the nominal RMS level is around -18 dB. That’s how we know that the maximum gain reduction we’ll need is 12 dB, and gives us a threshold target of something below -18 dB.
To set the threshold, right-click on the node to open the filter settings window and enter -18 in the “threshold” box. The quickest way to do that is to double-click on the box and enter a value on the keyboard, although you can also use the mouse to enter any parameter.
Whether or not the initial -18 dB threshold yields enough compression will have to be determined by ear. Lowering the threshold will increase compression. In practice, I usually end up lowering the threshold quite a bit from the nominal level, sometimes all the way down to silence. Resonances can require a LOT of compression.
This explanation for positioning a filter for resonance mitigation may have sounded like a lengthy process, but it really isn’t. It takes far longer to explain it than to do it. I’ve been doing it this way for a couple of years now, and it typically takes me less than a minute to set it up. Of course, there’s going to be the occasional extra-challenging track that takes a great deal more time and effort, but believe me, it would be a lot more work if you didn’t have something like MDynamicEQ on hand.
Using MDynamicEQ #2: Ducking a Full Mix from the Vocal
Adding an external sidechain to a compressor can turn it into a creative effect. Adding it to a multi-band compressor makes it a surgical effect. Being able to do that with a dynamic equalizer with all of MDynamicEQ’s options gives you a dizzying number of both creative and corrective possibilities.
Here’s a simple application: duck a specific band of frequencies in the instrument bus in response to the lead vocal. Granted, there are tools specifically designed to do this, but it’s a good excuse to examine MDynamicEQ’s sidechain features. Plus it actually works quite well for this purpose, especially for a dense rock mix where the challenge is a clear vocal while not raising the vocal as far above the instruments as you would for a pop tune.
The setup for this technique requires that all the instruments we want to duck are routed to a single bus, into which we’ll insert an instance of MDynamicEQ. Then we’ll create aux sends from each of the vocal tracks that we want to drive the ducking effect. Often, that’ll be just the lead vocal.
The idea is that by lowering the band of frequencies that most competes with the vocal, we’ll create a spectral space in the mix for the vocal. This allows us greater clarity from the vocal without having to actually turn it up, or to lower the instruments across the board to the point where they go all wimpy.
After routing each vocal track to the equalizer’s sidechain input, we’ll next want to identify which frequencies we want to duck. MDynamic’s sonogram feature is great for this, because it thoughtfully gives us the option to view the sidechain signal rather than the main audio.
To access this feature, click on the Settings button in the main edit window and select “Side Chain” as the Source. [Note: prior to version 8, there was a large button labeled “Side Chain” in the main window, but this has been moved to the Settings dialog to tidy up the UI.]
Here we see that the vocal’s fundamental frequency is around 200 Hz, but there are significant harmonics at 400 and 600 Hz.
What we could do is place three filters at 200, 400 and 600 Hz and tell them to compress those frequencies whenever the vocal is active.
However, we can get results faster and easier (and potentially more transparently) by using a band-shelf filter that spans the fundamental and the first one or two significant harmonics.
As noted earlier, the band-shelf is a cross between a bandpass and a shelf, reducing (or increasing) the gain uniformly over the specified range of frequencies, as opposed to the normal bell-shaped filter.
In the screenshot below, we’ve applied a single band-shelf filter that encompasses the 200 to 400 Hz range.
We could stretch this out to also include the 600Hz harmonic, or even up to the 800Hz component. For maximum transparency, it’s generally best to use the narrowest filter that works. Start with just enough width to cover the fundamental frequency, widen it until the effect becomes noticeable, then back it off.
In order for the ducking to not be overly obvious, we’ll have to be careful with the dynamics parameters. If the attack is too slow we might miss the leading edge of a vocal phrase; if too fast the effect will be obvious and distracting. The same applies to release times.
After positioning the filter(s), right-click on each filter being used and click on the “Sidechain” button in its Dynamics section. This exposes the filter to the sidechain signal.
One more tip…I’ll sometimes add a second filter in the upper midrange and duck those frequencies, too. That’s where the consonants live, which are the key to lyric intelligibility. If I find that the words are hard to understand, ducking the upper mids will usually clear them right up.
This ducking trick can be amazingly transparent, clarifying the vocal without actually turning it up and without losing instrumental impact. You can use it in anything from a simple ballad to a dense hard rock mix. It works so well for me that I’ve created a preset tailored to my own voice. Now, whenever I’m mixing a project in which I’m the vocalist, I can easily set up the ducking effect with just a few clicks.
Using MDyamicEQ #3: Kick and Bass Spectral Conflict Mitigation
This is a variation on the previous scenario, but one that’s more routinely employed: preventing spectral collisions between kick drum and bass guitar that cause the bass to mask the drum.
Traditionally, this is accomplished with a conventional broadband compressor or gate, with the kick drum driving the sidechain input on the bass track to duck the bass when the drum is struck.
The problem with the traditional approach is that the bass is completely ducked by the kick, making it tricky to set up in such a way that the effect isn’t glaringly obvious. Obvious may not always be what you want. For a less-intrusive, more transparent effect it would be nice to reduce only those frequencies that the kick and bass have in common, a task well-suited to a dynamic equalizer.
Setup is easy: insert an instance of MDynamicEQ in the bass track and route the kick drum to its sidechain input.
Follow a similar procedure to the one described in the previous section, using the sonogram to determine where the kick and bass frequencies overlap most. Place a bell, low-shelf or band-shelf filter over those frequencies.
I prefer a bell-shaped filter because it’s usually not necessary to duck all frequencies, just the fundamental frequency of the kick drum. Depending on the instrument tones and envelopes, there may be no need to duck the bass when it’s playing high notes that don’t actually interfere with the kick.
The filter’s Attack and Release settings are critical in this scenario, an example of when the “Auto” values might not be best. You’ll generally want a fast attack, between 1 and 3 milliseconds depending on the sharpness of the kick drum’s attack. The idea is to start the gain reduction as soon as the kick’s envelope begins so that you hear the drum first rather than the bass.
The release time needs to be short enough that there is no gap between the drum’s decay and the bass coming back in. The exact time depends on the kick drum’s envelope.
Here’s an example of a kick drum waveform, showing times for the attack and sustain portions.
The initial impact is 36 milliseconds in duration, about 25% of the total length of the drum sound.
We’ll probably want at least the first 36 milliseconds of the bass to be overridden by the kick sound, although every track is different and how much of the kick takes precedence will depend on the nature of the effect we’re after. For some genres, such as EDM, it might be appropriate to duck the bass for the full 146 milliseconds or even longer, while a classic rock or MOR pop song might do better with just the initial attack.
For this example, we’ll assume that transparency is the goal, so we’re mostly interested in making the attack phase stand out. We’ll also assume that we want the bass to start fading back in just as the drum fades out. This requires a short release time, perhaps around 10 milliseconds as a starting point and then lengthening it until it sounds right. Once we’ve gotten the release time just right, the kick and bass will act as if they were one instrument, with the kick providing the attack and the bass providing the body.
Yes, I’ve waffled on the exact values here, but that’s because there are variables: the envelopes of the kick and bass, the amount of effect we want, and what’s appropriate for a given style of music. In the end, you have to tweak it by ear, but it always helps to start with a mental image of the instruments’ envelopes and an idea of the timeframe.