/ˈfɪltər/
Filter is a signal processing tool that attenuates or removes specific frequency ranges from an audio signal. Defined by a cutoff frequency and slope, filters are used in synthesis, mixing, sound design, and mastering to sculpt tone and control spectral balance.
Every frequency you hear is a choice — and the filter is how you enforce it.
In music production, a filter is any circuit or algorithm that selectively attenuates or passes specific frequency regions of an audio signal. Unlike a fixed equalizer band, a filter is defined by a cutoff frequency — the point at which the signal begins to roll off — and a slope, expressed in decibels per octave, that determines how aggressively frequencies beyond that cutoff are reduced. The result is a shaped spectrum: some frequencies pass through largely unaffected, others are gently or severely diminished. That simple mechanism is responsible for the warmth of a vintage synthesizer, the thump of a kick drum cleared of rumble, and the sweeping tension of a filtered breakdown in electronic music.
Filters exist at every stage of the production chain. In synthesis, they are one of the three core building blocks alongside the oscillator and amplifier, and the filter stage in a subtractive synthesizer is where raw, harmonically rich waveforms are sculpted into recognizable timbres. In mixing, high-pass filters clean low-end congestion from non-bass elements, while low-pass filters tame harshness. In mastering, gentle shelving filters and steep brickwall low-pass filters at 20 kHz or 21 kHz enforce the bandwidth of a final delivery file. In sound design, automated filter sweeps create movement and drama that no static EQ can replicate.
The most fundamental filter types are the low-pass filter (LPF), which passes frequencies below the cutoff and attenuates those above; the high-pass filter (HPF), which does the opposite; the band-pass filter (BPF), which passes a range of frequencies centered around a center frequency while attenuating both below and above; and the notch filter (band-reject filter), which attenuates a narrow band while leaving surrounding frequencies intact. Each type has distinct musical applications and sonic character.
The parameter that gives filters their character beyond simple attenuation is resonance (sometimes called Q or emphasis). By feeding a portion of the filter's output back into its input at the cutoff frequency, resonance creates a peak of boosted energy just at the cutoff point. At subtle levels resonance adds presence and color; at moderate levels it creates the classic synthesizer filter sound; at extreme levels it causes the filter to self-oscillate, generating a pure sine tone even without an input signal — a behavior exploited extensively in acid house, techno, and experimental music.
Understanding filters is not optional for serious producers. They are present in every EQ plugin, every synthesizer, every dynamic processor with a sidechain, and every de-esser. The vocabulary of cutoff, slope, resonance, and filter type applies uniformly across hardware and software contexts. A producer who understands filters deeply can diagnose mix problems by ear, program synthesizer patches with intention, and make automation choices that serve the music rather than obscuring it.
At its core, a filter is a frequency-dependent network that changes the phase and amplitude of different parts of a signal. In analog circuits, filters are built from resistors, capacitors, and inductors (or operational amplifiers with reactive components) arranged to create voltage dividers whose impedance varies with frequency. The simplest first-order RC low-pass filter rolls off at 6 dB per octave above its cutoff frequency — defined as the point where the signal is attenuated by approximately 3 dB. Cascading multiple filter stages multiplies the slope: two stages yield 12 dB/oct, four stages 24 dB/oct, the classic slope of the Moog transistor ladder filter.
In digital audio, filters are implemented as difference equations operating on discrete sample values. A Finite Impulse Response (FIR) filter computes each output sample as a weighted sum of a fixed number of past input samples, guaranteeing linear phase — meaning all frequencies are delayed by the same amount, preserving transient integrity. An Infinite Impulse Response (IIR) filter feeds previous output samples back into its calculation, allowing steep slopes with far fewer coefficients and therefore lower CPU cost, but introducing phase shift that varies with frequency. Most synthesizer filters and many mixing EQs use IIR designs modeled on analog prototypes; linear-phase EQ plugins used in mastering typically use FIR designs.
The cutoff frequency (fc) is the primary control parameter. In synthesis contexts it is often modulated by envelopes, LFOs, or MIDI velocity to create time-varying timbral change — this modulation is the defining character of synthesizer filter sweeps. The resonance parameter creates a boost at fc by introducing positive feedback inside the filter structure. In the Moog ladder topology, this feedback is introduced between the output of the fourth stage and the input of the first; in the state-variable filter topology used in the Roland Juno, Oberheim SEM, and many modern designs, separate outputs provide simultaneous LP, HP, and BP responses from the same circuit. The state-variable design is particularly valued for its stable, musical resonance behavior at high Q settings.
Filter slope (also called order or roll-off) determines how sharply frequencies are attenuated beyond the cutoff. A 6 dB/oct (first-order) slope is gentle and musical, often used for high-pass filtering of room mics or adding warmth to a mix bus. A 12 dB/oct (second-order) slope is the most common in synthesizers and general mixing. A 24 dB/oct (fourth-order) slope, as in the Moog ladder, creates powerful, dramatic timbral control. Steeper slopes of 48 dB/oct or higher appear in brickwall limiting and mastering contexts where absolute bandwidth control is required. The steeper the slope, the more pronounced the phase shift introduced by IIR designs — a real trade-off to weigh in mastering or mix bus work.
In practice, the interaction of cutoff, resonance, and slope defines the entire sonic personality of a filter. A 24 dB/oct Moog-style filter at 50% resonance sweeping from 200 Hz to 4 kHz sounds nothing like a 12 dB/oct state-variable filter sweeping the same range — the topology, the feedback structure, and the non-linearities introduced by analog saturation inside the filter circuit all contribute to character that producers learn to identify and seek out deliberately.
Diagram — Filter: Frequency response curves for low-pass, high-pass, band-pass, and notch filters, showing cutoff frequency, resonance peak, and roll-off slope.
Every filter — hardware or plugin — operates on the same core parameters. Know these and you can work with any implementation.
The cutoff frequency (fc) marks the -3 dB point of the filter's response — frequencies above this point in a low-pass filter are progressively reduced; frequencies below it pass through. In practice, sweeping the cutoff is the primary performance gesture of filter-based synthesis, ranging from a few hertz to the upper limit of audible range (20 kHz). Automating cutoff on a synth pad from 300 Hz to 3 kHz over 8 bars creates one of electronic music's most recognizable tension-and-release shapes.
Resonance, also labeled Q or Emphasis, boosts a narrow band of frequencies centered on the cutoff point by feeding a fraction of the filter output back into its input. At moderate settings (30–60%) it adds the characteristic 'singing' quality of classic analog filters; at high settings (80–100%) it produces dramatic, almost vocal tonal color. Beyond a topology-dependent threshold, resonance causes self-oscillation — the filter generates a sine wave at fc with no input signal, a technique used to create pitched bass stabs and effects in acid house (Roland TB-303) and techno.
Slope describes how aggressively the filter attenuates frequencies beyond fc. Common values are 6 dB/oct (first-order, gentle), 12 dB/oct (second-order, most common in mixing EQ), 24 dB/oct (fourth-order, classic Moog character), and 48 dB/oct or steeper (brickwall, mastering). Steeper slopes offer more isolation but introduce more phase shift in IIR designs. Choosing 12 dB/oct for a high-pass on a vocal preserves natural low-mid body, while 24 dB/oct cuts harder and may thin the sound if the cutoff is set too aggressively.
The four primary filter types — low-pass, high-pass, band-pass, and notch — define the fundamental topology of the filter's response. Many synthesizers and multimode filter plugins allow switching between types or blending them continuously (morphing filters). Choosing a band-pass instead of a low-pass for a synth lead gives it a hollower, more mid-forward sound; switching to high-pass creates an airy, telephone-like quality. In mixing, the choice of filter type determines whether the intervention is additive (BP boost) or subtractive (HP or LP cut).
In synthesizers, the filter envelope amount controls how far (and in what direction) an ADSR envelope pushes the cutoff frequency over time. A high positive envelope amount with a fast attack and medium decay produces the classic 'wah' pluck of a Minimoog or TB-303 bass. A negative envelope amount inverts the response — the filter closes quickly and opens slowly — producing a reverse-swell character. This parameter is absent from static mixing EQs but is central to any synth patch or creative filter effect built with an LFO or envelope follower.
Key tracking (keyboard tracking or keyboard follow) causes the cutoff frequency to rise and fall in proportion to the pitch of the MIDI note being played. At 100% key tracking, the harmonic structure of the filter's effect remains perceptually constant across the keyboard — a patch won't become muddier in lower octaves and brighter in higher ones. At 0%, the cutoff stays fixed regardless of pitch. Partial key tracking (50%) is common for giving a synth pad subtle timbral variation across the range without fully tracking pitch.
Session-ready starting points. These values are starting points for in-context adjustment — trust your ears and the specific recording over any table.
| Parameter | General | Drums | Vocals | Bass / Keys | Bus / Master |
|---|---|---|---|---|---|
| HPF Cutoff (cleaning) | 80–120 Hz | 20–40 Hz (kick), 100–150 Hz (snare room) | 100–150 Hz | 30–60 Hz (bass), 80–100 Hz (keys) | 20–30 Hz |
| LPF Cutoff (air control) | 12–18 kHz | 10–14 kHz (OH room bus) | 14–18 kHz | 4–8 kHz (bass), 14–18 kHz (keys) | 20–21 kHz (brickwall) |
| Resonance (mixing) | 0–10% | 0–5% (transient clarity) | 0–8% | 0–12% | 0% (flat, transparent) |
| Slope (HPF) | 12 dB/oct | 12–24 dB/oct | 12 dB/oct | 6–12 dB/oct (gentle on bass) | 6–12 dB/oct |
| Notch depth (hum removal) | 30–60 dB at 50/60 Hz | 20–40 dB at resonant freq | 30–60 dB at 50 Hz or sibilance | 20–40 dB at resonance | 18–30 dB (surgical) |
| Synth filter cutoff (starting point) | 800 Hz–2 kHz | 600 Hz–1.5 kHz (perc synth) | N/A | 200–600 Hz (bass), 1–3 kHz (lead) | N/A |
| Synth resonance (musical) | 20–50% | 10–30% | N/A | 30–70% (bass), 20–50% (lead) | N/A |
These values are starting points for in-context adjustment — trust your ears and the specific recording over any table.
The history of the audio filter is inseparable from the history of electrical engineering. The mathematics of reactive circuits capable of frequency-selective behavior were established by Oliver Heaviside and others in the late 19th century, but it was Campbell and Wagner's independent invention of the electrical filter in 1915 — primarily for telephone multiplex transmission — that established the practical filter as a communications tool. By the 1940s, active RC filter designs using vacuum tubes allowed audio engineers to build equalizers for broadcast and recording. The Pultec EQP-1, introduced in 1951 by Eugene Shenk and Ollie Summerland at Pulse Techniques, used passive shelving filter networks followed by a tube amplifier stage to create a highly musical, phase-coherent EQ that remains one of the most emulated filter topologies in plugin history.
The decisive moment for the filter as a creative musical tool came in 1964 when Robert Moog, working from Harald Bode's earlier voltage-controlled amplifier research and inspired by composer Herbert Deutsch, developed the transistor ladder filter — a four-pole (24 dB/oct) low-pass design using a cascade of transistor pairs in a ladder configuration with global negative feedback to implement resonance. The Moog filter's distinctive warmth came from transistor non-linearities and soft saturation inside the feedback path. Its sound was immediately exploited by Wendy Carlos on Switched-On Bach (1968), reaching mainstream ears and establishing the synthesizer filter sweep as a legitimate compositional device. The Minimoog Model D (1970) put the ladder filter into a portable, affordable instrument that defined the sound of progressive rock, funk, and eventually hip-hop bass.
The late 1970s and 1980s brought competing filter topologies that each produced distinct characters. Oberheim's SEM (1974) used a two-pole state-variable filter designed by Tom Oberheim and engineer Dave Rossum, offering simultaneous LP/HP/BP outputs and a gentler, more open resonance character favored for pads. Roland's IR3109 and CEM3320 chips, used in the Juno-106 (1984) and Jupiter-8 (1981), produced a cleaner, less saturated sound suited to pop. The Roland TB-303 Bass Line (1981), initially a commercial failure, used a 24 dB/oct diode-ladder filter that — when Yasushi Saito's original acid-house artists like DJ Pierre and Spanky discovered the unit second-hand around 1985–1986 — produced the squelching, resonating bass lines that defined acid house and later techno.
The DAW era shifted filter implementation to software, beginning with early plug-in formats in the mid-1990s. Steinberg's VST specification (1996) enabled third-party developers to model analog filter topologies digitally. FabFilter's Pro-Q series (Pro-Q 1, 2009; Pro-Q 2, 2013; Pro-Q 3, 2018) advanced digital filter design by offering zero-latency and linear-phase modes, per-band channel linking, and mid-side filtering in a single interface that became the industry standard. Simultaneously, the boutique hardware market saw renewed interest in analog filter modules — particularly in the Eurorack format popularized by Dieter Doepfer from 1995 onward — preserving the character of transistor ladder, OTA, and diode-ladder designs in modular systems that could be integrated with DAW-based production workflows.
In mixing and arrangement: The high-pass filter is arguably the most-used tool in any mix session. Applied to every non-bass element — guitars, pads, room mics, backing vocals, pianos — a 12 dB/oct HPF set between 80 and 150 Hz removes energy that doesn't belong to those instruments while freeing up headroom for kick drum and bass. Engineers like Chris Lord-Alge and Tchad Blake have spoken extensively about the high-pass filter as the first move on any channel: a 100 Hz HPF on a rhythm guitar with a 12 dB/oct slope is invisible in solo but contributes enormously to mix clarity. The complementary move — a gentle low-pass filter on a full mix to remove ultrasonic energy above 18–20 kHz — prevents intermodulation artifacts and reduces limiter stress on the master bus.
In synthesis and sound design: The filter is the primary tone-shaping tool of subtractive synthesis. A producer building a bass patch on a Moog Subsequent 37 or its software equivalent starts with a sawtooth wave rich in harmonics, then uses the filter cutoff to decide which harmonics survive. A low cutoff (around 200–300 Hz) produces a clean, round sub; as the cutoff rises past 1 kHz the bass acquires growl; past 3–4 kHz it becomes aggressive and distorted-sounding. Adding an envelope with a fast attack and moderate decay to the filter cutoff replicates the physical behavior of acoustic instruments, where higher harmonics decay faster than fundamentals — giving the synthesized sound an organic, plucked quality. Velocity-to-cutoff modulation makes the patch respond expressively to playing dynamics.
In sidechaining and dynamic filtering: Filters combine with dynamics processors to create frequency-selective control. A sidechain high-pass filter on a compressor — available in Waves SSL G-Channel, UAD SSL 4000 E, and most modern channel strips — tells the compressor to ignore low-frequency energy below a set threshold, typically 80–100 Hz, when calculating gain reduction. This prevents kick drum transients from pumping the vocal compressor. Similarly, a de-esser is fundamentally a band-pass sidechain filter connected to a compressor or dynamic attenuator targeting sibilance frequencies (typically 5–10 kHz on female vocals, 6–12 kHz on male).
In arrangement and automation: Filter automation is a structural technique in electronic music. The extended filter sweep — automating a low-pass filter's cutoff from closed (100–200 Hz, essentially muting all harmonic content) to fully open over 4, 8, or 16 bars — creates tension and anticipation that resolves at the drop. Producers including Aphex Twin, Daft Punk (Bangalter and de Homem-Christo), and virtually every techno producer use LPF automation as a compositional device equivalent to a crescendo in orchestral writing. In contrast, a high-pass filter sweeping upward from the sub bass can hollow out a section, removing weight and creating a breakdown feel without silence.
One email a week. The techniques behind the terms — curated by working producers, not algorithms.
Abstract knowledge becomes practical when you can hear it in music you know. These tracks demonstrate filter used intentionally, at specific moments, for specific purposes.
The defining example of a resonant low-pass filter in popular music. The entire track is built around a Korg MS-20 (or similar analog) bass line filtered aggressively with high resonance, creating the characteristic squelching, nasal tone. Listen to the first four bars: the bass line is barely identifiable as a bass — it's more of a pitched, growling filter artifact. The cutoff sweeps slightly during verses, fully opening only during the hook to reveal the underlying harmonic content. The decision to leave the filter nearly closed for most of the track creates a sense of compression and release that works as arrangement architecture.
The opening harp-like loop is a sampled instrument passed through a gentle low-pass filter that removes any high-frequency bite, leaving only warmth and fundamental harmonics. This application demonstrates subtractive filtering in a sample-based context: the original source material is too bright and present, but the filtered version sits inside the ambient texture. At 0:28, when Elizabeth Fraser's vocal enters, the contrast with the filtered loop demonstrates how dynamic filtering — keeping the instruments spectrally narrow — creates space for unprocessed vocal clarity. The sub-bass underpinning has its own HPF applied above 30 Hz to prevent subsonic rumble from the original recording.
The synth bass line in the main body of the track is a master class in filter envelope programming. The cutoff envelope has a near-instant attack, a decay of roughly 200–300 ms, and a sustain that leaves the filter partially open — giving each note a sharp transient click at the attack, a bright mid-section, and a warm tail. At high resonance the filter self-oscillates at pitch, blurring the boundary between filter artifact and pitched instrument. James's use of the filter here illustrates how envelope amount, decay time, and resonance interact to define the rhythmic and melodic character of a bass line entirely through filtering, with the oscillator waveform acting merely as a carrier.
The opening piano sample has been aggressively high-pass filtered to remove everything below approximately 300–400 Hz, leaving only the upper harmonics and transient attack of each note. This transforms what would be a full, warm piano into a thin, percussive, almost toy-piano texture that sits in the upper-mid pocket of the arrangement. The choice creates room for the eventual 808 sub to dominate the low end without competition. This is mixing-as-filtering: the piano's spectral footprint has been deliberately sculpted to serve an arrangement role rather than preserve the instrument's natural timbre.
The foundational document of filter-as-composition. The entire track is built from TB-303 bass lines with cutoff and resonance adjusted manually during performance — a workflow that became the template for acid house and later techno. The 303's diode-ladder filter at high resonance creates a sound that moves between musical pitch, noise, and pure filter artifact depending on cutoff position. Listen for the moment around 2:30 when the cutoff opens sharply and resonance backs off — the underlying note content of the bass line becomes audible for the first time. That reveal is a pure filter technique applied as structural composition.
The low-pass filter passes frequencies below the cutoff and attenuates those above it. It is the most common filter type in both synthesis and mixing, used to remove high-frequency harshness, create warm tones, and generate the classic filter sweep effect. In mixing, HPF cleaning and LPF air control on individual channels and buses are standard workflow steps. The character of a low-pass filter is defined almost entirely by its topology: ladder, state-variable, diode-ladder, and Sallen-Key designs each produce distinctly different timbral signatures.
The high-pass filter passes frequencies above the cutoff and attenuates those below, making it the primary tool for removing low-end rumble, room resonance, and low-frequency buildup in mixing. On most analog console channel strips the HPF is a fixed 18 dB/oct or 12 dB/oct circuit with a sweepable frequency knob. In mixing practice, applying an HPF to every non-bass element below 100 Hz (and to bass elements below 30–40 Hz) is considered fundamental hygiene. Aggressive HPF settings with resonance boosts transform the tool into a creative instrument, hollowing out sounds for electronic textures.
The band-pass filter passes a range of frequencies centered on a center frequency (fc) while attenuating both lower and higher frequencies on either side. Q (resonance) determines the width of the pass band: high Q produces a narrow, telephone-like resonant peak; low Q creates a broad, gentle mid-range emphasis. Band-pass filters are essential in formant synthesis, vowel-sound emulation, and the creation of vocal-like textures in synthesizers. In mixing, a narrow band-pass sweep automation on a synth layer during a breakdown creates movement without using reverb or delay.
The notch filter, also called a band-reject filter, attenuates a narrow frequency band while leaving surrounding frequencies intact. The primary mixing use is surgical removal of problem frequencies: 50 Hz or 60 Hz hum from electrical interference, specific room resonances on drum recordings, or feedback rings on live recordings. Notch filters are also used creatively in phaser effects, where a comb of notches is swept across the frequency spectrum, and in acoustic guitar recording to remove the 'wolf tone' resonance of the instrument's body. A steep, deep notch (30–60 dB attenuation) at a single frequency is often invisible to listeners even in solo.
The ladder filter is a specific circuit topology, not a frequency response type. Four cascaded transistor (or diode) stages create a 24 dB/oct low-pass response with a feedback path from output to input implementing resonance. The non-linear saturation of the transistors gives the ladder filter its characteristic warmth and the perception that it 'breathes' at high resonance settings. At maximum resonance the ladder self-oscillates cleanly and musically. The transistor ladder is considered the gold standard of analog filter warmth and is the most-emulated topology in digital filter plugins.
The state-variable filter simultaneously produces low-pass, high-pass, band-pass, and notch outputs from the same circuit, allowing the filter type to be selected or continuously morphed. Its resonance characteristic is smooth and stable even at very high Q values, making it preferred for pads and melodic content where the ladder's aggressive self-oscillation would be too intense. The Oberheim SEM's state-variable filter is particularly prized for its ability to stay musical at near-maximum resonance settings. Many modern filter plugins implement state-variable topology for precisely this stability.
Frequency conflicts — two instruments in the same range at similar levels — are the root cause of muddy mixes.
These MPW articles put filter into practice — specific techniques, real tools, and applied workflows.