|
 |
|
|
|
|
Understanding filtering
For our next music tech tutorial, RA's Jono Buchanan offers a straight-forward guide to the world of filters.
Here's a quiz question: what do sieves, coin slot machines and low bridges have in common? If you're wondering how such a question could have anything to do with music production, let me explain. Each of those things is designed to allow certain things to pass through its assorted restrictions, while trapping or preventing the passing through of other objects. Any of these objects might be a useful image to have in your mind as you begin the process of filtering; the audio equivalent. Filters, within synthesizers, as effects plugins or even within EQ plugins do precisely the same—allowing some elements of the audio frequency spectrum to be heard and others to be trapped, or, as we're talking about audio, dramatically reduced in volume.
However, we need to take a step further back to really appreciate the capabilities and uses of filter effects by understanding that first and foremost our ears are filters themselves. There are frequencies which exist which are below the range of the average person's hearing, while there are plenty more which exist above our upper hearing thresholds too. While we're all different and indeed our hearing capabilities change as we grow older, the average person can hear from 20Hz (cycles per second) at the bass end to 20,000Hz, or 20kHz at the top. This wide range allows us to hear everything from rumbling tube trains a station down the line to the most ear-piercing screechy treble, explaining in part why our ears are drawn to music which provides a combination of these frequencies, and plenty more in between.
However, due to the nature of multi-track recording and mixing, you'll be aware a mix won't work well without a balance across the audible frequency bands. If you want to add extra brightness to a sound within a mix, you'd turn to an EQ effect and make the necessary adjustment, while the same might be true in reverse—if a sound has too much high-end, you'd accordingly dial some treble down in volume. Like EQ effects, filters are used to control the tone of a sound but there are some fundamental differences between the two types of effect and this article will explain them, as well as looking at some practical uses of filtering effects.
For starters, EQ effects tend to feature multiple bands, which can be independently adjusted up or down—cut or boost, in other words. You could use a single EQ plugin to roll out bass, boost low mid-range, scoop out upper mid-range and enhance treble for instance, simply by adjusting the bands of your EQ to the relevant frequencies and then boosting or cutting as you wish.
Unlike filters, EQs can change tone across several frequency bands at once.
Filters don't do this. While they too allow you to select a frequency point at which they'll go to work, the frequencies outside of their range will be dropped in volume. This may happen fairly subtly or much more extremely but the important point is that filters aren't used to boost volume as their name suggests; they cut, or filter out unwanted frequency content. In another striking difference from EQ effects, filters only feature a single band of operation. They allow you to choose a single frequency to act as the point at which filtering will start and then drop the volume of frequencies beyond that frequency. As you'll know if you've ever spent time with a synthesizer or a filter effect, this point is called the "cutoff" frequency, which makes sense, as frequencies beyond this are indeed dramatically reduced in volume.
Most filter effects are what are called multi-mode, as they're capable of operating in a variety of ways. The most common filter type is called a "low-pass filter" (often abbreviated to LPF) which, as its name suggests, allows low frequencies to pass through and be heard, while cutting off upper-frequency content. The effect of using a low-pass filter effect on a sound is to make its tone deeper, darker and muddier. A naturally-occurring example of a low-pass filter is a wall. Imagine standing in a queue outside a club—all you'll hear is thumping bass, the only frequency group with enough energy to permeate the wall. Whenever the door to the club opens, much higher frequency content will rush out to meet you as the filtering process offered by the wall will have been temporarily removed. When the door closes again, the wall will go back to being an effective low-pass filter.
We can replicate this naturally-occurring effect with a filter placed in the output channel of a project. By putting a filter there, it will affect every part of the mix, so it's an effective way to achieve the global filtering approach achieved naturally by a wall. To do this, place a filter in the output channel and select the low-pass filter mode if your filter plug-in offers multiple options. Then bring the cutoff frequency down towards the bottom. As you do, you'll hear the brightness being sucked out of the mix, leaving you just with murky low-end content. Whenever you move the cutoff point back up, you'll regain the upper frequency content and the mix will become brighter, as you can hear here.
Whole track LPF filtering: As you can hear, the mix plays through at full-frequency until bar 8 when the low-pass filter's cutoff then sweeps down. Through the remaining 8 bars, it creeps back up.
The next most-common filter type is a high-pass filter (HPF), which does precisely the inverse of a low-pass filter. It too uses the cutoff point to determine the frequency at which filtering will occur but rather than letting low frequencies through to be heard, it lets high frequencies through instead. Correspondingly, it will drop the level of frequencies below the cutoff point. This, as you might expect, leaves a very thin sound, particularly when the cutoff frequency is set high enough only to leave sparkling treble frequencies, cutting mid-range as well as bass.
One of the reasons high-pass filters sound so unnatural (even more so than low-pass filters) is that if you apply the same technique we heard in the previous example—putting a high-pass filter across an entire mix—for many of the sounds present, only their harmonics remain. Recorded and synthesized sounds tend to be rich in harmonics, with a loud fundamental frequency at the bottom and a series of related frequencies, called harmonics, present in a sound as well, stretching upwards. The reason for the unnatural quality is that no natural sound can produce such upper harmonics without a fundamental frequency present to trigger them, yet a high-pass filter will retain the sound of the harmonics while reducing the volume of the fundamental. As this simply isn't possible in the real world of natural acoustics, high-pass filters bring an exciting, un-real option to the filtering table as you can hear in this clip.
Whole track HPF filtering: This time a high-pass filter is used and as you can see from the automation ramp, it teases you into the second half of the track by dipping once and re-rising, before dipping again to allow the bass through properly.
The third most common filter type is called a band-pass filter (BPF). This sets a band of frequencies around the cutoff point and lets them through to be heard, cutting out bass frequencies below and treble frequencies above. This is a sound we're all familiar with because of telephones. If you're put on hold during a call and you're played music, you'll notice that both the bass and the treble frequencies are reduced, leaving sufficient frequencies to hear the music, albeit in a fairly compromised way. Band-pass filter effects can be created using a low-pass filter and a high-pass filter in tandem but band-pass filters achieve the same result without having to set up two separate filters.
Whole track BPF filtering: Finally, here's a band-pass filter working on the same mix. As only a narrow band of frequencies is let through around the cutoff point, this clip starts like a high-pass filter and sounds like a low-pass filter when it drops. But only a band-pass filter can create this narrow effect.
So now we've heard examples of the three most common filter types working across a whole mix but, of course, filters aren't only used in this way. The three main components of a synthesizer—the oscillator, filter and amplifier sections—are responsible for pitch, tone and volume control, so filters play an integral part in shaping the tone of individual sounds too. When you look at the filter section of a synthesizer, as well as being able to select the filter mode you want to use (if multi-modes are available), the other dominant dial alongside cutoff will be resonance. The dictionary definition of resonance in this context would be something like "resonance boosts the volume of the frequency at the cutoff point."
In case that doesn't make musical sense right away, think about it like this: We already know that, regardless of which filter mode you're working with, the cutoff point acts as a threshold, preserving the volume of frequencies on one side, while cutting the volume of frequencies on the other. The cutoff point has to be "somewhere" therefore, and resonance's job is to turn up the volume of frequencies regardless of where cutoff is placed. Again, this is unnatural. Usually, the further you get away from the fundamental frequency, the quieter harmonics get, yet with a low-pass filter, you could place the cutoff somewhere around the upper harmonics, where they're dying out in volume, then use resonance to dramatically boost their volumes at this point. This gives an edge and bite to the sound as you can hear in the examples below.
Filter sweep: Here you can hear a rich, synthesizer sound using a low-pass filter's cutoff frequency to sweep down from top to bottom through the frequency range. There's no resonance in this sweep.
Filter sweep 40% resonance: Here's the same sound with 40% resonance. You can hear that there's more bite in the sweep now, as each group of frequencies is being boosted in volume as the cutoff point sweeps down.
Filter sweep 65% resonance: Finally, here's the same sound with 65% resonance. That bite in the sound is even more exaggerated now; you can easily pick out the sweeping harmonics this time.
Envelopes also play a huge part in determining the behaviour of filters. Remember, envelopes allow you to change parameters over time, so any envelope allied to the filter section will change a sound's tone over time. Even though the previous audio examples were created using automation of the cutoff point, they could just have easily been created using envelopes, as the movement was "ramp-shaped" from the top of the frequency range to the bottom. Most synthesizer envelopes are four-staged affairs, with controls for attack, decay, sustain and release. It's easiest to imagine envelopes affecting volume, as here it's easy to imagine attack time as a fade in, release as a fade out and so on. However, as we're discussing filters, rather than an envelope affecting volume, here you'll be affecting tone with sounds getting brighter or duller as cutoff is affected by the four stages of movement.
In the previous audio examples, the tone "speaks" immediately and this would be achieved by setting a very quick, even immediate attack time. The downwards ramp would be created via the decay time which, in our examples, is fairly long. The sustain level will be the point where the cutoff point comes to rest after the decay stage, so if your cutoff point is set towards the top, you'll struggle to hear decay at all as there's nowhere for the cutoff to go between the immediate attack and the high sustain value. The release time will affect what happens to the tone of the sound once you let go of the notes you're holding and, as such, is hugely dependent on the volume envelope stage. If you set a long filter release time, you'd rightly expect to hear the tone of the sound change as you let go of the keys you're holding. However, if the volume envelope's release is set to a very short time, the filter release simply won't have time to be heard before the volume is chopped off the sound. So, to hear a longer filter release time, you'll need at least as long a volume envelope release time.
EXS24 Envelope routing: Below you can see how envelope 1 is set up to control the filter cutoff, with the ADSR faders at the bottom, the target and source routing in the middle and the filter cutoff, resonance and other settings at the top.
You can hear how crucial filter envelope shaping is in the following examples, where I've created a bass sound to suit the beat loop it's working alongside.
Bass too fat: This sound doesn't really sound sufficiently bass-y, as the cutoff frequency is set too high, so too many frequencies are getting through.
Bass too subby: By dropping the cutoff point, the low-pass filter is more obviously doing its job but now there's not quite enough bite at the start of the sound—the bass is getting lost.
Bass bite: Now, with an appropriate decay time and a little of the envelope (number 2 in our case) applied to the cutoff frequency, the filter bites nicely at the start of each note.
Interestingly, inverting filter shapes can work really well too. As you might expect, if an envelope is inverted, rather than the decay acting as a "bite" which moves rapidly down to the sustain level, instead it will "suck" up to the sustain level instead. This is a great way to get basslines and kick drums working well together as any kicks/bass notes which fall at the same point won't overload with bass frequency. Instead, the bass note will start as an almost imperceptibly quiet sub-bass rumble and then get brighter. With a little trial and error, it's perfectly possible to get the rise to fit musically with the speed of your track.
Sucky bass: You can hear how the bass rises up to the cutoff frequency at the start of each note. This is still set by filter envelope decay but with an inverted (upside down) envelope. You can hear it by itself first, then with the beats.
LFOs are also hugely important tools for getting filters moving, as anyone who enjoys "wobble" dubstep patterns will testify. LFOs are like regular oscillator shapes which inflict their shape and speed upon other parts of a signal. Using a sine wave shape which rises and falls, you can force the filter cutoff to correspondingly rise and fall so that the tone of the sound opens and closes. Simply set up an LFO with the speed and shape of your choice and then use the routing options within your synth to patch this LFO into the filter Cutoff.
MorphoX LFO routing: By the cursor, you can see how LFO1 (on the GUI's left) routes into the filter cutoff.
LFO Bass: Here an LFO clocked to tempo shifts the filter cutoff in time with the beats.
The LFO amount will prove crucial too—too much and tone shift will be too wide, too little and it will barely be audible. If you're using a plugin synth and the LFO speed can be clocked to the tempo of your project, this movement can make musical sense too, as the speed of movement will be in time with the rest of your track. If you want multiple speeds so that the rate of movement can change from one note to the next, either automate the LFO speed so that it changes for each note, or set up multiple versions of the same sound with different LFO speeds from one to the next. Then, when you've played in the notes you want, chop them up and move them to the assorted tracks you've created so that each note plays back at a different speed.
LFO multi-speed bass: Here the same sound is used with different LFO speeds across three tracks. The bassline is chopped with each region moved down to the track and speed of choice.
You might be wondering how extremely a filter reduces the volume of content beyond the cutoff point. Again, this is often variable with an example setting of 12dB/octave. What this means is that for each octave of harmonics beyond the cutoff point, volume will be reduced by 12dB. For more extreme filtering, increase this number; for more subtle filtering, reduce it.
EXS24 filter strengths: Logic's EXS24 filter features four "strengths" for its low-pass filter (24, 18, 12 and 6dB per octave) and a fixed 12dB per octave for its HPF and BPF options.
If you're looking on in envy wishing that such treatments were possible for non-synthesizer sounds (such as vocal parts or anything else you've recorded), worry not: all of this capability can be applied via filter plugins native to your DAW. In fact, this is exactly how the first audio examples were created, by using a filter plug-in (UAD's Moog Filter) inserted into the output channel of the mix. The parameters are the same—you'll have a choice over which filter mode you use, there'll be cutoff and resonance controls, plus envelope and LFO options. So dive in and enjoy the wonderful world of filtering as creatively as you can.
|
|
|
|
|
|
Published / Wednesday, 24 October 2012
 15 Comments
|
|
|
|
|
|
|
Features
|
|
|
