Engineering Foundations · Evidence-backed reference

Understand the system beneath the setting.

A processor number becomes useful only when you understand the signal, detector, room or source that gives it meaning. This reference connects the engineering model to the audible decision—and makes the limits explicit.

6 deep chapters 3 live calculators 9 primary references 1 controlled method

Use the numbers correctly

Three small calculators for large mistakes.

These tools expose scale: how quickly phase changes, how sample offsets depend on sample rate and where a rectangular room’s simplest resonances may occur. They inform a listening test; they do not choose the mix.

Cycle and offset

Milliseconds to phase angle

See how the same delay represents a different part of a cycle at each frequency.

Sample alignment

Samples to milliseconds

Convert a DAW nudge into time at the session’s actual sample rate.

Ideal rectangular model

First axial room modes

Estimate the first three length, width and height modes using 343 m/s as the speed of sound.

01

Filters, phase and transient integrity

An EQ curve describes magnitude. The time behaviour depends on the filter design, slope, Q, frequency and surrounding signal.

Immediate answer

Use the least complex filter mode that performs the audible job. A normal causal EQ is usually a sensible starting point for a kick. Choose linear phase only when preserving inter-channel or parallel magnitude relationships outweighs its latency and possible pre-ringing. Judge the first 20–50 ms as carefully as the spectrum.

Magnitude

How strongly each frequency is passed or attenuated. This is the curve most EQ interfaces foreground.

Phase

How sinusoidal components are shifted relative to one another. Frequency-dependent phase produces frequency-dependent delay.

Impulse response

How the filter behaves in time. It reveals ringing, latency and whether energy appears before or after a sharp event.

Minimum phase is not “phase-free”

A typical low-latency parametric EQ changes magnitude and phase together. Around a boost, cut or cutoff, different frequencies shift by different amounts. That can alter the peak shape of a kick or the way it sums with another layer. It does not automatically make the sound worse. Analogue filters also have phase response, and many familiar records contain it.

“Minimum phase” describes a system that realises a given magnitude response with the smallest compatible group delay among stable causal systems. It does not mean minimum audible consequence, and the exact result depends on the filter. A gentle bell far above a kick’s fundamental and a steep high-pass near it are not equivalent merely because both use the same mode.

Linear phase moves the compromise

A linear-phase filter delays frequency components by an equal amount, preserving their relative phase through the filter. Implementations require latency, and a symmetrical finite impulse response can distribute ringing before and after a transient. On a sustained signal this may be unobtrusive. On a sparse low-frequency hit, pre-ringing can soften the sense that energy begins at one instant.

Higher low-frequency resolution requires a longer impulse response. That is why some linear-phase processors offer resolution or latency choices: low frequencies need more time to be represented precisely. Linear phase can be useful for certain crossovers, parallel paths or mastering moves, but it is not an automatic “quality” switch.

SituationStart withWhyVerify
Ordinary tonal correction on one kickNormal low-latency modeDirect workflow with no pre-event ringingLevel-match and inspect onset plus body
High-pass near the fundamentalGentler slope or source envelope firstSteep filtering can reshape time and resonanceCompare at the same low-frequency loudness
Parallel path that must recombineMatched filters on both paths or a deliberate linear-phase testUnmatched phase can create unintended combingNull, mono and polarity tests
Occasional resonanceDynamic EQ or source editA static notch changes every hitCheck detector timing and false triggers

Why steep slopes are expensive

A sharper transition demands a more selective filter. In a causal minimum-phase design that generally increases phase rotation and group-delay variation near the transition. In a linear-phase design it generally demands a longer impulse response, increasing latency and ringing duration. If a brickwall filter seems necessary to separate two kick layers, the layers may not have distinct roles.

Controlled filter test

  1. Match bypass loudness without auto-gain.
  2. Loop the busiest kick–bass passage.
  3. Listen to the onset, then the tail, then their combined groove.
  4. Switch filter mode without moving the curve.
  5. Keep the mode whose side effect is least important in context.
02

What a compressor actually detects

A compressor is a nonlinear, time-dependent system. Threshold, ratio, attack and release do not fully specify its behaviour.

Immediate answer

Choose a compressor by the envelope behaviour you need, not its label. Detector type decides what counts as “level”; topology decides where that level is measured; timing laws decide how gain moves; knee and gain computer decide how firmly the change begins. Two compressors at the same displayed settings can react differently.

The signal path has two jobs

The audio path carries the sound whose gain will change. The sidechain or detector path estimates level and tells the gain element what to do. Some compressors expose the sidechain; others hide it. A detector high-pass filter does not remove sub from the audible kick. It reduces how strongly deep energy controls gain reduction.

Feedforward designs measure the input and calculate the desired output reduction. Feedback designs measure after the gain element, so the control action depends partly on its own result. Neither word guarantees a sound, but the topology changes the relationship between threshold, ratio, timing and programme material.

Design choiceWhat it followsPossible kick consequence
Peak detectorShort excursionsCatches onset spikes readily; can react strongly to a narrow first cycle
RMS-like detectorEnergy averaged over timeOften follows body more than the very shortest click
Hard kneeRatio changes abruptly around thresholdMore distinct boundary between untouched and compressed hits
Soft kneeRatio increases gradually around thresholdSmoother onset of control across varied velocities
LookaheadDelays audio so the detector can act before a peakReliable peak capture with latency and possible transient softening
Sidechain high-passReduces deep bass contribution to detectionLess pumping from the kick tail; more onset may pass

Attack is not a stopwatch placed on the waveform

Attack normally describes how the control envelope approaches gain reduction after the detector crosses into compression. Manufacturers can define this using different time constants and reference points. The displayed value therefore predicts direction more reliably than identical behaviour across devices.

On a kick, a longer attack often allows more onset through, but only if the detector, threshold and source create enough time before substantial reduction. A very high peak can drive the control circuit quickly even with a nominally slower setting. Conversely, an RMS detector may respond later because its estimate already averages over time.

Release can distort low frequencies

A 50 Hz sine cycle lasts 20 ms. If gain changes substantially within individual low-frequency cycles, the waveform is amplitude-modulated rather than simply made quieter. Very fast attack or release can add harmonic distortion; a longer release can hold the body or next kick down. Auto-release systems analyse programme behaviour and use more than one time constant, which is why one “auto” mode can feel different from another.

Choose by job

  • Peak containment: high threshold, fast controlled action, most hits untouched.
  • Body consistency: detector and timing that follow sustained energy without erasing onset.
  • Onset softening: deliberately faster action, level-matched against bypass.
  • Density: low-ratio or parallel control, with tail and noise monitored.

Controlled compressor test

  1. Remove makeup gain.
  2. Set the threshold from the amount of reduction, not a memorised number.
  3. Sweep attack through its full useful range.
  4. Set release against repeated hits and bass notes.
  5. Restore equal loudness and decide in the full arrangement.
03

Polarity, phase, delay and rotation

These terms describe related but different operations. Treating them as synonyms leads to unreliable layer and microphone decisions.

Immediate answer

Use polarity inversion as a two-state comparison, delay as a time relationship, and phase rotation as a frequency-dependent timing change. None of them can globally “align” two kicks whose pitch and envelopes evolve differently. Define which region—onset, body or tail—must combine coherently.

OperationMathematical ideaUseful question
Polarity inversionMultiply every sample by −1Does the opposite sign improve the important overlap?
Constant delayShift the entire signal in timeDo audible arrivals become one event?
Phase rotationChange phase by frequency, often with an all-pass networkCan peak shape or band interaction improve without the same magnitude change?
Envelope trimLimit the duration of interactionCan the second layer stop competing after its job is complete?

cycle period (ms) = 1000 ÷ frequency (Hz)  ·  phase angle (degrees) = delay ÷ period × 360

At 50 Hz, one cycle lasts 20 ms, so a 4 ms offset represents 72 degrees. At 100 Hz, the same offset represents 144 degrees. A single delay therefore cannot give the same phase relationship across the kick’s spectrum. With a descending pitch envelope, the relationship also changes during the hit.

Layer alignment is a role problem

If two layers both contain full-range onset, body and sub, they create several simultaneous alignment problems. Make one layer the anchor and restrict the other to a missing function. A high-passed click layer mainly needs onset coherence. A sub layer needs stable low-frequency addition and may tolerate a different visible transient position.

Align by the first meaningful acoustic or electronic arrival, not automatically the file boundary or largest peak. A sample can contain leading silence, noise or a gradual pressure build. Use the waveform to find candidates, then use the full pattern to decide.

Multiple microphones add acoustic delay

An inside kick microphone, outside microphone, overheads and room microphones receive the event at different times and from different acoustic paths. Moving a microphone changes tone, bleed and phase together. A digital nudge can align one feature, but it does not recreate the spatial relationship of another placement.

The familiar 3:1 placement guideline is a bleed-management starting point, not a guarantee of phase coherence. In recording, capture the best individual tones and relationships first; correct later only when the correction improves the intended image and groove.

Why a correlation meter is insufficient

Correlation summarises similarity over a window. A positive value does not prove the kick’s low body sums well, and a negative moment does not prove an intentional stereo transient is wrong. Inspect the relevant time and band, then verify mono at matched level.

Controlled alignment test

  1. Choose the anchor and name the added layer’s job.
  2. Restrict the added layer to that job.
  3. Compare polarity without changing level.
  4. Nudge in both directions around the audible arrival.
  5. Check repeated hits, mono and the loudest arrangement.
04

Low-frequency acoustics, modes and monitoring

The kick arriving at your ears is the source, loudspeaker, room and listening position acting together.

Immediate answer

If a narrow low-frequency problem changes when you move through the room, treat it first as monitoring evidence—not immediate permission to EQ the kick. Reposition monitors or listener, measure more than one point, control decay acoustically where possible, and use calibration mainly to reduce peaks rather than force deep nulls upward.

wavelength (m) = speed of sound (approximately 343 m/s) ÷ frequency (Hz)

A 50 Hz wavelength is about 6.86 metres. In a small room, the distance between surfaces is comparable to fractions of these wavelengths, so reflected energy forms strong position-dependent patterns. A frequency can be loud at a pressure maximum and nearly absent at a null. Turning that frequency up in the mix may compensate at one seat while making every other system excessively heavy.

Room modes have frequency and decay

Axial modes involve one pair of opposing surfaces; tangential and oblique modes involve more surfaces. The calculator above lists only the simplest axial modes of an ideal rectangular room. Real doors, windows, construction, furniture and connected spaces change the measured response.

A frequency-response graph shows level at the measurement window. It does not by itself show how long energy persists. A kick can sound “boomy” because a mode decays slowly even when the average curve is not spectacularly high. Waterfall, spectrogram or decay views help separate excess duration from static level.

Speaker-boundary interference is not the same as a mode

Direct sound from a monitor combines with a delayed reflection from a nearby wall, floor or desk. At frequencies where the paths oppose, a cancellation appears. Moving the monitor or listener changes the path length and therefore the cancellation frequency. Deep geometric nulls are poor targets for large corrective boosts because the extra energy can also cancel at the same position while consuming headroom and excursion.

EvidenceLikely categoryFirst moveAvoid
Level changes sharply when listener moves forwardRoom mode or boundary cancellationTest position before mix EQBoosting the kick from one null
Low note remains after the kick stopsModal decay, source tail or bothMute playback and compare room decay with the waveformAssuming every boom is in the file
Both channels together lose energy near one bandSpeaker/listener geometry or sub crossoverMeasure speakers separately and togetherCorrecting both channels identically without diagnosis
Sub is loud but poorly definedLevel, crossover, phase, placement or decayCalibrate level and crossover, then inspect time responseUsing sub level as a substitute for extension

Calibration cannot repeal acoustics

Room correction is valuable for level, delay and correctable response errors. It can reduce modal peaks and integrate a subwoofer. It cannot make an untreated room position-independent, restore headroom lost to a deep cancellation or shorten a room’s physical decay in the same way as effective treatment.

Monitor at one repeatable working level, check quietly, compare trusted references and walk the room when a bass judgement seems extreme. Small position changes are diagnostic tools. Genelec’s placement guidance likewise emphasises symmetry, reflection paths and moving monitors or listener when a mode or null dominates.

Loudspeaker excursion

Deep bass requires large cone displacement. Long kick tails and subsonic content can consume excursion and amplifier headroom without adding useful audible identity. A high-pass filter can protect a system, but its cutoff and slope must be chosen with the kick’s fundamental and phase behaviour in mind.

Controlled room check

  1. Play a trusted reference and the problem mix at matched level.
  2. Move the listener forward and backward.
  3. Measure left, right and combined response.
  4. Inspect decay as well as magnitude.
  5. Correct placement and level before deep mix EQ.
05

Measurement, loudness and true peak

Meters answer specific questions. Problems begin when a valid number is used as evidence for a different question.

Immediate answer

Use sample peak for digital sample headroom, true peak for reconstructed peak risk, loudness for programme-level comparison, spectrum for frequency distribution and time-domain views for envelope and alignment. None of them measures “punch” directly. Let the listening question choose the meter.

MeasureUseful forDoes not prove
Sample peakLargest stored sample valueMaximum reconstructed analogue waveform level
True peakEstimated intersample peak through oversampled reconstructionPerceived loudness or kick impact
RMS / energy averageSignal energy over a selected windowEqual perception across frequency and duration
LUFS / LKFSK-weighted programme loudness under a defined algorithmThe correct loudness target for every creative master
Crest factorRelationship between peak and average levelWhether the transient sounds musically punchy
Spectrum / spectrogramDistribution of energy by frequency and timeCause, masking or quality without context

Why true peak can exceed sample peak

Digital samples describe points from which a continuous waveform is reconstructed. The curve between stored samples can rise above every sample value. Filtering, sample-rate conversion and lossy encoding can also create higher peaks. ITU-R BS.1770 specifies an algorithm for programme loudness and true-peak estimation; the true-peak section exists because stored sample maxima are not the entire reconstruction story.

True-peak margin is delivery risk management, not a tonal prescription. A lower true peak does not automatically mean a better kick. If controlling it audibly blunts the source, reconsider the onset, clipping strategy, oversampling or delivery margin rather than treating one ceiling as universal.

Loudness is programme measurement

Integrated loudness describes a programme over time with gating; short-term and momentary views use shorter windows. A single kick is too brief to interpret like a finished programme, and a kick’s K-weighted loudness value is not a reliable ranking of impact. Use loudness to level-match passages or versions, then listen to envelope, groove and spectrum.

EBU R 128 defines a broadcast loudness-normalisation framework. Streaming services and music releases can use different operational targets. Kickopedia therefore treats standards as measurement definitions and delivery contexts, not universal creative targets.

Spectrum settings change the answer

An analyser trades time resolution against frequency resolution. A longer FFT window separates low frequencies more precisely but averages over more of the changing kick. A shorter window shows onset timing more clearly but spreads low-frequency bins. Window function, smoothing, averaging and display slope all alter the picture.

For a descending kick, a static peak may represent the resting tail, a resonance or an average across the sweep. Use a spectrogram to see movement, then confirm with controlled filtering or resynthesis. Do not notch every visible ridge.

A defensible A/B comparison

  1. Align versions in time.
  2. Match perceived or defined programme loudness.
  3. Randomise or hide identity when practical.
  4. Use the same playback path and level.
  5. Repeat after a pause and in the full arrangement.

Export diagnosis

  1. Render lossless at the session rate.
  2. Re-import with warping and normalisation off.
  3. Align and compare through the same master path.
  4. Check true peak before and after sample-rate conversion.
  5. Test the delivery codec only after lossless render matches.
06

Source physics, synthesis and acoustic recording

Kick sounds are created by different systems. Their parameters are not interchangeable, even when the results share an adjective.

Immediate answer

Identify the source architecture before prescribing a fix. An acoustic membrane, resonant analogue circuit, sampled drum machine, pitch-envelope oscillator, FM network and physical model distribute control differently. Edit the parameter closest to the cause: excitation, resonance, pitch trajectory, damping, sample playback or microphone position.

SourcePrimary controlsCharacteristic strengthCommon failure
Acoustic kickHeads, tuning, beater, damping, shell, performance, microphones, roomComplex onset and coupled resonancesTrying to EQ away a tuning, damping or placement problem
Analogue resonant circuitExcitation, resonance, decay, pitch/tone, nonlinearitiesContinuous, coupled responseAssuming every long electronic kick is an 808 circuit
Sample playbackFile choice, start, pitch, interpolation, envelope, layeringImmediate access to captured detailPitching also changes duration or moves the onset unnaturally
Pitch-envelope oscillatorWaveform, start/end pitch, sweep curve, amplitude decay, click/noiseDirect control of electronic kick anatomyA disconnected click or audible sine tail with no body transition
FM / phase modulationCarrier, modulator, index, ratios and envelopesEvolving harmonics and compact metallic attackHigh-frequency density becoming unstable or overly pitched
Physical modelExciter, membrane/body parameters, damping, coupling, radiationInterdependent performance-like controlTreating one parameter as a simple EQ band

Acoustic kick: fix the instrument first

A kick drum couples a struck batter head, enclosed air, resonant head, shell and surrounding room. Beater material and impact point change the onset. Head tension and damping change modes and decay. A pillow or other damping reduces sustain and overtones; it also changes feel and low-frequency development. Record a useful instrument before asking processors to reconstruct one.

An inside microphone aimed toward the beater generally emphasises attack and isolation. Moving outward or toward the resonant head captures a rounder, more developed low-frequency event with more room and bleed. Shure’s guidance presents these as starting relationships and explicitly recommends placement experiments. The microphone’s polar pattern, proximity effect, maximum SPL behaviour and mounting also matter.

Two microphones create a blend and a timing problem

An inside microphone can supply onset while an outside microphone supplies body. Their levels should be set for roles before alignment. Flip polarity as a comparison, then adjust placement or delay while listening to the intended band. Include overheads: a close-mic combination that sounds huge alone may weaken the kit when overhead timing is restored.

Do not automatically time-align every microphone to the closest one. Natural arrival differences communicate size and space. Correct a relationship when it sounds unfocused, hollow or rhythmically late—not because the waveforms are visually different.

808 and 909 are architectures, not adjective bundles

Roland describes the TR-808 bass drum as an analogue synthesis abstraction built around a sweeping sine-like low-frequency event with adjustable decay. Its identity comes from the coupled circuit behaviour and pitch movement, not merely “a sine wave plus distortion.” A modern 808 bass often means a longer, pitched descendant used as a bass voice; that production role is broader than the original machine.

Roland describes the TR-909 as a hybrid instrument overall: analogue drum synthesis with digital recordings for cymbals and hi-hats. The kick belongs to its analogue drum-synthesis side and presents a different, more assertive attack/body relationship. “808 versus 909” should therefore compare envelope and architecture in a musical role, not reduce the machines to one frequency each.

FM and physical modelling fill different gaps

FM or phase-modulation synthesis is useful when a simple pitch envelope cannot create the desired body complexity. A short modulation-index envelope can produce a dense onset that relaxes into a more stable tail. Integer and non-integer frequency relationships lead to different harmonic or inharmonic structures. Because several envelopes interact, level-match changes carefully.

Physical modelling starts with an exciter and a model of a vibrating system. For drums, membrane modes, damping, coupling and collision behaviour can produce responses that evolve as connected physical variables. It is valuable for controllable acoustic plausibility and unusual hybrids, but the model is an abstraction whose accuracy depends on assumptions and numerical design.

Source-first selection

  1. Name the kick’s rhythmic and low-end job.
  2. Choose an architecture whose envelope naturally fits it.
  3. Set pitch trajectory before static EQ.
  4. Set damping or decay in the full pattern.
  5. Add only the missing onset, body or sub function.

Common category errors

  • Calling any long sine bass an “808” without distinguishing instrument, synthesis and role.
  • Using sample pitch to fix an envelope problem.
  • Using EQ to compensate for acoustic microphone placement.
  • Layering full sources before assigning roles.
  • Presenting artist-associated reconstructions as documented studio facts.

Primary and technical references

Equations and calculators are deliberately idealised. Real processors, rooms, instruments and listening systems must be measured and heard in their actual context.

Apply the foundation

Move from model to diagnosis.

Use the decision guides when you know the system and need to choose the smallest useful production move.

Open decision guides