Filters & Common Traits

A Dry/Wet knob controls whether you hear none of the effect or nothing but the effect.

Technically an "effect" can be any operation or process that chances a sound in some very general way. It might add entirely new sounds on top of an old, it might just change existing properties of an existing sound. A "filter" strictly applies pass/deny rules to sound properties as explained below.

But whether or not the effect you're using, probably via a plug-in of some kind, is a filter or not there will be some common terminology. For instance it's important to discern the difference between the "level" or "indluence" of an effect/plug-in and the "Dry/Wet" control you may see within the knob itself.

Pass filters

A filter simply describes a way to remove various qualities of a signal, whether it be the frequency, amplitude, phase or a combination of any of these inherent qualities. The best way to understand what is meant by that is to examine actual examples. Here are the most common linear filter effects you might work with.

For each of the four filter passes below, the target frequency was about 600hz with a transition period of about 200hz.

Visual approximation Audio Sample Description
Low-pass
Frequencies on the low end of the range are allowed to pass through.
High-pass
Higher frequencies are allowed to pass through, low are muted.
Band-pass
Frequencies within a certain width are retained while everything above and below that width are muted.
Notch
Mutes the signal of a given frequency range and allows everything above and below to pass through.
The cutoff of a filter is the exact fequency at which the amplitude reaches zero.

There are other examples of filtering signals with various titles. A "comb" filter for instance works like a band pass but has several frequency ranges spaced apart that it lets pass. In the waveform view the "bandform" might them have the appearance of a comb.

Some might call a filter that modifies only the phase of a signal, but leaves frequency amplitude untouched, an "All-pass" filter. Though you'll usually just find such a thing labeled with the actual words "phase" or "phaser".

The Transition band of a filter is the frequency range between the natural amplitude and the cutoff of a passing filter.

For each filter, the width of the transition band is important to consider, since it is present in almost every filter type. This describes the speed and velocity of the filtering. For instance if a band pass starts at 1000 and you've set the cutoff at 1500 then the transition band has a width of 500.

It's possible to have a transition of zero and have no observable transition. For scientific applications this can be desirable. For musical engineering however, it often produces a harsh, un-organic sound.

Q

The "Q", or "quality" value of a filter is usually akin to the transition band of the filter. Different plug-ins will operate in slightly different ways but for the most part the Q dial will adjust the speed at which the pass goes from full signal to no signal.

Equalizers

Probably the most recognizable form of signal processing is the equalizer. While not as prevelent as it was during the great boom box era of the 80's and 90's, equalizers can still be found in most playback devices and are therefore more recognizeable than most filter interfaces.

As you've probably guessed an equalizer is essentially a set of notch filters set to predefined bands. Instead of adjusting an amplitude and frequency in a seperate filter effect for each desired range the user simple adjusts a single amplitude for each "percieved" range.

That last part is the key of why equalizers are fairly easy to use. The sliders (or knobs) of an equalizer typically do not apply to the same widths. By that we mean while the range for the first slider might be 20 to 40, the second might be double that width, and extend from 40 to 80, and so on. In a normal 10 band equalizer you might notice that the mid range, frequencies between approximately 100 to 2000 or so, have sliders that cover only about a 100 to several hundred each. The upper range sliders on the other hand might cover a frequency range of several thousand.

This is because it is easier for the human ear to percieve changes in certain ranges than others. Most people will have no problem telling the difference betwenn 100 Hz and 200 Hz. When you get into the super low, or higher ranges, then it becomes more difficult. The difference between 5000 Hz and 5100 Hz is almost impossible to hear.

Not all filters work by reducing or boosting certain frequency ranges though.

Low frequency oscilation

Modulation: Using one waveform to change the properties of another without introducing a discernable sound from the first.

Low frequency oscillators (LFO) typically control bands that are at the low end or beneath the audible range. But while they themselves can not be heard, they can be used to help manipulate other sounds by applying their own waveforms as a modulator. If a LFO has a speed of two cycles a second, then applying those cycles to the volume will make the amplitude of the audio waver two times a second.

If you're familiar with stringed instruments you can think of using an LFO as similar to Vibrato. This is the technique where a musician will physically vibrate their entire hand while holding down a string to the neck of an instrument to produce a less constant and more "energetic" tone.

Limiters & Gates

Limiters and noise gates are often included in the same plug-in or effect because they both deal with adjusting the amplitude or gain of a sound. You can think of them as similar to low and high band passes but instead of affecting the frequency of a sound they affect the amplitude.

Limiters

Limiters prevent clipping by lowering the signal amplitude below the maximum possible strength.

A limiter exists to prevent clipping which might otherwise flatten an audio track to a single waveform, destryoing any detail or in a file or individual track. Typically it is applied to the higher amplitudes of a track.

Within limiter plug-ins the primary distinction is whether or not they are "hard" or "soft" limiters.

A "hard" limiter will simply cut off all amplitude above a given amount.

A "soft" limiter will try and "curve" the signal strength as it reaches a maximum limit so that the cut off is not so sudden by interpolating it from the last acceptable level. This means that if a signal exceeds the chosen limit the samples around that area will be brought up or down so that the audio wave leads into the maximum value slowly instead of simply snapping to it.

Noise Gates

A Noise gate drops signals below a given threshold to remove quieter sounds while retaining higher.

Inversely, Noise Gates are used to prevent sound below a given level from getting through. This helps prevent background static and noise but can reduce the subtleties of recorded sounds like breathing or light instruments.

A noise gate can be thought of as a filter that affects amplitude instead of frequency. Typically it is applied to the lower amplitudes of a track.

Shared controls

  • Gain - Similar to gain in other effects. It is usually included to make up for any widespread ceiling application.
  • Ceiling - This defines the amplitude level where sound will be clipped.
  • An ADSR envelope. The best way to define how clipped samples interact with samples before and after is through a simple ADSR envelope. It works the same way and can apply to either before or within the cut area depending on the plug-in. For instance, a longer attack will make the lead into the clipped area smoother and a long release will make the lead out from the clipped area smoother, but both will then quiet the surrounding sounds.

The controls have similar functions when it comes to noise gates. A gate is either closed (not allowing the sound through) or open (allowing it) so the questio is how do we decide what gets through and what doesn't.

  • Gain - the amplitude level the audio has to exceed to be considered (typically left at 0, but can be raised to leave very low sounds.
  • Threshold - The amplitude level at which the gate will no longer be applied.

In other words, if the amplitude level falls between the gain and the threshold, it will be clipped to the last known acceptable level.

Limiters are especially useful with audio that you want to have a quick uptake or downturn. In the same way that a "ceiling" setting on a limiter can shave off the top of an audio wave, the noise gate can be used to shave off the sides of a wave form. This can be used for instance in a heavy metal song where you want a particularly "chunky" sound for drums or guitars that should enter quickly and leave almost as fast.