DIY crossovers and multiband processing in your DAW
By Phil O'Keefe
Crossovers are not normally something people think of when it comes to DAW software. Many musicians will be familiar with them as part of a home or car stereo system, bass amp and speaker rig, PA / DJ system, or keyboard setup. We'll get to how you can use them to do some useful tasks in your DAW in a moment, but for those who need a refresher…
A crossover is a passive or active device that uses equalization filters to divide the frequency spectrum into multiple bands, and then allows you to patch and route each band individually. A two band crossover "splits" the signal into separate low and high frequency ranges. The point where the signal "splits" is called the crossover frequency. How quickly it transitions or "rolls off" frequencies beyond the crossover point is called the filter slope, and is usually measured in decibels per octave. A 12 dB per octave filter will reduce the level by 12 decibels for every octave past the crossover cutoff frequency, while a 24 dB per octave filter will be twice as steep, and signals one octave past the cutoff frequency will be attenuated 24 decibels.
Crossovers are commonly used in things like bass amplification and PA systems; allowing different parts of the frequency spectrum to be routed to separate power amplifiers, along with speakers that are optimized for each band. Multiband (also called bi-amped, tri-amped, two-way, three-way, etc.) setups generally offer better fidelity than a full-range speaker system, since speakers can be optimized to work in narrower frequency ranges instead of trying to reproduce the entire audio range. The amount of amplification can also be optimized for each frequency band and the connected speakers.
CROSSOVERS IN YOUR DAW
However, in recording, there are also times when you might want to divide the frequency spectrum of a sound into different components. I recorded some basic audio clips and took lots of screen shots to help demonstrate the sound of multiband processing and show you how to set it up. A basic mono bass part (Fig. 1 and audio Clip 1 - clips are at the bottom of the article) is our test subject, although the principles could be just as easily applied to other types of sounds. It was tracked in Pro Tools over a simple drum beat generated with FXPansion's BFD Eco. From there, it was manipulated in a few different ways, which I'll describe as we go along.
Figure 1: A basic mixer layout with a mono bass track
Let's say you'd like to use the mono bass recording from example 1 and add some distortion to it, but when you insert a distortion as an inline plugin (Fig. 2), the bottom of the mix gets muddy, or the fuzz plugin causes the lows to practically disappear completely as can be heard on Clip 2. Wouldn't it be nice to be able to add the "dirt" just to the upper frequencies of the bass signal, without having to process and mess up the bottom end?
Figure 2: A distortion plugin inserted inline on a track affects the entire sound at all frequencies
Or maybe you'd like to add a little phaser or vibrato to the bass on the slow part of that new song you're working on, but when you insert the plugin as a channel insert, the effect is too overblown; making the whole track sound too wobbly. (Fig. 3 / Clip 3) With simple tools such as track cloning, or DAW mixer aux sends and return channels, and basic plugin EQ filters, you can do plenty of frequency splitting, and once you've split it up into different bands, each can be processed individually and mixed together to create a composite sound.
Figure 3: A phaser-vibrato plugin running inline as a channel insert will modulate all frequencies
Sure, there are some cool multiband distortion and compression plugins out there, but unfortunately, there are not many "crossover" plugins that allow you set the crossover frequencies and the type of filters and their cutoff slopes, and also set the number of bands and assign each to a separate mixer output channel so you can process them individually. However, there are a couple of easy ways you can set up similar processing with the tools your already have built into your DAW.
What if we could split a sound -- any sound in your DAW -- into multiple channels, each covering a different frequency range? Then we could process each "slice" or range of frequencies independently before mixing everything back together again. That would be pretty cool, huh? By splitting a signal into multiple bands, you can do all sorts of things to it from there, such as widen the apparent stereo width or subtly chorus it in one frequency range while distorting it or using a sub bass synth plugin in another. It also allows you to EQ and compress each frequency range independently.
METHOD ONE - CLONE AND FILTER
One way to do it, assuming you have a fast enough hard drive to handle the extra streaming, is to just clone the track you want to "split up" and process the original and the clone tracks differently with various EQ filters. The advantage here is that it's relatively easy to set up; just select a track and use your DAW's clone function to create copies of it. The downside is that this approach at least doubles the amount of audio data that the "track" needs to stream from the hard drive. If you want to use a 3-band crossover, it will triple it. Figure 4 (and Clip 4) show our original bass track along with a cloned version of it, and Figure 5 and Clip 5 add in a distortion plugin on the high frequency channel.
The original bass track has a low pass filter inserted with a 200Hz cutoff frequency; this becomes the "low frequency" channel. The clone of our bass track has a high-pass EQ filter (bottom EQ plugin in the image), with the cutoff also set to 200Hz. This essentially filters out all of the low frequencies, and turns this channel into the "high frequency" channel for processing and mix purposes. You can hear the sound of this "crossover" in Clip 4. Compare it to to original mono bass track (Clip 1) and see if you can notice any differences in the bass sound near the 200 Hz crossover frequency.
Figure 4: Low Pass EQ filtering on the original track, and high pass filtering on a clone of that track basically divides the track into two in a manner similar to a crossover
Figure 5: Inserting a distortion on the channel with the high-pass filter allows you to affect only the higher frequencies of the bass, while leaving the bottom alone
METHOD TWO - DIVIDE & CONQUER
An arguably better way to go, assuming you have a few aux sends and returns to spare (and most modern DAWs have plenty) is to set up two (or more) pre-fader aux sends to route the signal off of the original source track and bring it back into the mix on multiple but separate aux return channels. If you just want to separate the signal into two bands; high and low frequencies, two aux sends and two returns is all you need (Fig. 6 / Clip 6). If you want a three band "crossover", you would need three aux sends and returns, with a high-pass and low-pass filter for the high and low frequency bands, and a band-pass filter for the midrange channel. If you don't have a "bandpass" setting on your favorite EQ, you can use a lowpass and a highpass filter together to achieve the same thing; rolling off the highs and lows and leaving only the midrange. In Figure 7 (and Clip 7) a distortion plugin has been added to the aux return channel with the high pass filtering; again, this results in the bass signals above 200Hz being distorted, while those below that cutoff / crossover frequency are left unprocessed and without distortion, resulting in a much fuller and more solid sounding bottom, as can be heard in Clip 7.
Figure 6: You can use multiple pre-fader aux sends and aux return channels along with EQ filters to separate a sound into different bands. Don't forget to mute the source track!
Figure 7: A distortion plug in has been added to the aux return with the high pass filtering so that only the higher frequencies of the bass track will be distorted
If you wanted crossover frequencies at 150Hz and 1kHz, use a low-pass EQ filter set with a steep 18 or 24 dB per octave rolloff on one of the aux return channels. On the second return, use a bandpass EQ (or separate high and low pass filters) set for 18-24db per octave slopes to get rid of everything below 100Hz and above 1kHz (leaving you with a "midrange" band), and finally, for the third "high frequency" aux return channel, use a high-pass filter set at 1Khz. Feel free to experiment with different crossover frequencies until you find what works best. For example, the vibrato in Clip 8 (Figure 8) sounded better to me with a slightly higher crossover frequency of 500 Hz instead of the 200 Hz setting that I used with the distortion examples.
Figure 8: Don't be afraid to try different crossover frequencies. In this example, the crossover point has been moved up from 200 Hz to 500 Hz, which gave a more subtle vibrato effect
In general, I recommend picking a filter type and setting and sticking with it throughout the project - always using 18dB, or sticking with only 24db per octave slopes for the entire project and in general, I think the steeper the better -- but remember -- there's a lot of debate and preferences over different types of filters and different slopes in crossovers and filtering in general, so feel free to experiment with different settings and different EQ plugins until you find what you like the sound of.
There is a potential for some "weirdness" at the crossover frequencies. Sometimes you can get a bit of a volume boost or dip at the crossover frequency, or a bit of phase shift. Crossovers are always a compromise, but the creative and problem solving possibilities of multiband processing may be worth the slight tradeoff in absolute fidelity. Only your ears can answer that for you.
Don't be afraid to mix one band louder or softer than the others to get the sound you want. Things like distortion can change the harmonics and apparent loudness of a band, and sometimes you may need to attenuate that band a bit to compensate.
A mono to stereo plugin can be great to add some stereo width to a sound in one band, while leaving the other bands in mono. If you experiment with stereo effects processing on one or more band, make sure you check your mix for mono compatibility to insure that there is no phase cancellation happening.
Phil O'Keefe is a multi-instrumentalist, recording engineer / producer and the Senior Editor of Harmony Central. He has engineered, produced and performed on countless recording sessions in a diverse range of styles, with artists such as Alien Ant Farm, Jules Day, Voodoo Glow Skulls, John McGill, Michael Knott and Alexa's Wish. He is a former featured monthly columnist for EQ magazine, and his articles and product reviews have also appeared in Keyboard, Electronic Musician and Guitar Player magazines.
I've used the clone and filter method when doubling vocal tracks but I have crashed the program a few times when it got too processor intensive. I've got another computer coming in the end of the week. After I get things set up, I'll have to try the "divide and conquer" method you describe. Very helpful article.