Audio for AV/Lesson 12

Filters and EQ 11 min read

Filters and Equalization

Understand filter shapes, frequency, gain, bandwidth and Q, then separate microphone correction, loudspeaker voicing and room EQ.

LEARNING OBJECTIVES
  • Explain frequency, gain, bandwidth and Q.
  • Distinguish pass filters, parametric filters and shelving filters.
  • Compare parametric and graphic equalizers.
  • Place microphone, room and loudspeaker EQ correctly in a DSP file.
  • Recognize problems that equalization cannot solve.

Why an AV engineer should care

Equalization appears everywhere in a DSP:

  • Removing low-frequency rumble from microphones
  • Correcting a resonant lectern
  • Voicing loudspeakers
  • Building crossovers
  • Improving speech clarity
  • Controlling feedback frequencies
  • Making program audio sound natural in the room

It is also easy to abuse. A DSP file containing fifteen extreme filters may look impressively technical while quietly indicating that someone tried to correct bad microphone placement, poor acoustics and a rattling ceiling tile with one heroic EQ curve.

The objective is not to make the response graph look busy. The objective is to remove identifiable problems while preserving natural sound and adequate headroom.


1. What an equalizer changes

An equalizer changes signal level according to frequency.

Three basic questions define most EQ adjustments:

  1. Where? The frequency being affected.
  2. How much? The amount of boost or cut in decibels.
  3. How wide? The range of nearby frequencies affected.

A parametric filter might be described as:

  • Center frequency: 500 Hz
  • Gain: –4 dB
  • Q: 3

That means the filter applies its maximum 4 dB cut at 500 Hz and affects a relatively narrow range around it.

The equalizer does not know whether 500 Hz contains:

  • A legitimate part of someone’s voice
  • A loudspeaker resonance
  • A room reflection
  • HVAC noise
  • Feedback
  • A guy humming while waiting for the Teams call to start

It changes everything passing through that frequency range. The engineer must determine what the problem actually is.


2. High-pass and low-pass filters

High-pass filter

A high-pass filter, or HPF, allows frequencies above its cutoff region to pass while progressively attenuating lower frequencies.

It is also commonly called a low-cut filter.

Useful microphone applications include reducing:

  • Building vibration
  • Table bumps
  • Handling noise
  • HVAC rumble
  • Plosives
  • Unneeded low-frequency energy

Speech microphones usually do not need to reproduce subwoofer content. Removing useless low-frequency energy improves headroom and may help automixing and AEC.

Low-pass filter

A low-pass filter, or LPF, allows lower frequencies to pass while attenuating frequencies above its cutoff region.

It may also be called a high-cut filter.

Applications include:

  • Subwoofer crossover processing
  • Limiting high-frequency noise
  • Band-limiting special audio feeds
  • Protecting loudspeaker components
  • Preventing unwanted ultrasonic content from reaching later stages

Q-SYS defines its current HPF and LPF blocks in these same functional terms and provides several filter families, including Butterworth, Linkwitz-Riley and Bessel-Thomson. (help.qsys.com, help.qsys.com)

Slope

Pass filters do not create a vertical brick wall. Their attenuation increases gradually beyond the cutoff.

Common slopes include:

SlopeNominal stopband attenuation per octave
6 dB/octave6 dB
12 dB/octave12 dB
18 dB/octave18 dB
24 dB/octave24 dB
48 dB/octave48 dB

A 12 dB/octave HPF approaches 12 dB of additional attenuation for each octave farther into its stopband. Attenuation near cutoff depends on filter family and order.

Steeper is not automatically better. Steeper filters create more phase rotation and may interact differently around crossover regions.


3. Parametric EQ

A parametric equalizer gives independent control over:

  • Center frequency
  • Gain
  • Bandwidth or Q

Q-SYS currently allows frequency, gain, bandwidth and Q control and supports parametric, low-shelf and high-shelf bands. Biamp similarly defines its parametric EQ using center frequency, gain and bandwidth. (tesira-help.biamp.com, help.qsys.com)

Frequency

The center frequency is where the filter applies its maximum boost or cut.

Gain

Gain determines how far the filter moves the response:

  • Positive gain boosts.
  • Negative gain cuts.
  • Zero means the band does nothing.

Q and bandwidth

Both describe filter width.

  • Higher Q: Narrower filter
  • Lower Q: Wider filter
  • Wider bandwidth: Lower Q
  • Narrower bandwidth: Higher Q

A simplified relationship is:

Q=(center frequency) / (bandwidth in Hz)

Bandwidth may instead be expressed in octaves. A one-octave parametric filter corresponds to approximately Q 1.41, though exact interpretation can vary slightly among filter implementations. Shure documents Q and width as two ways of representing the same control: increasing Q narrows the affected range. (pubs.shure.com, shure.com)

Typical uses:

AdjustmentTypical shape
Correct broad tonal imbalanceWide, low-Q filter
Reduce a room or device resonanceMedium-width cut
Attenuate a feedback frequencyNarrow, high-Q cut
Add general warmth or presenceGentle, wide adjustment
Make twelve random boosts because the graph looked lonelyPlease don’t

4. Shelving filters

A low-shelf filter boosts or cuts the frequency range below its selected turnover frequency.

A high-shelf filter does the same above its selected frequency.

Unlike a parametric bell, the response does not return to zero after passing through the affected region. It levels out into a shelf.

Useful applications include:

  • Gently reducing excessive low-frequency warmth
  • Adding or reducing overall high-frequency brightness
  • Compensating for a source with a broad tonal imbalance
  • Creating a user-facing bass or treble control

Shelves are useful for broad adjustments. They are generally poor tools for correcting one narrow resonance.


5. Graphic versus parametric EQ

A graphic equalizer divides the audio range into fixed center frequencies and fixed widths. The engineer adjusts only the gain of each band.

Common configurations include:

  • One octave
  • Two-thirds octave
  • One-third octave
  • One-sixth octave

Current Q-SYS graphic EQ components range from six bands at two octaves per band to 61 bands at one-sixth octave per band. Biamp provides one-third, two-thirds and one-octave configurations. (tesira-help.biamp.com, help.qsys.com)

A parametric equalizer lets you choose the precise center frequency and width.

Graphic EQParametric EQ
Fixed frequenciesAdjustable frequencies
Fixed band widthsAdjustable bandwidth or Q
Fast broad tonal changesPrecise corrective work
Familiar fader interfaceRequires more understanding
May affect more spectrum than necessaryCan isolate a problem narrowly

In modern installed DSP systems, parametric EQ is usually the better engineering tool. Graphic EQ remains useful for broad live adjustments or a familiar operator interface.

Shure describes the parametric EQ as the problem-solving tool and the graphic EQ as the broader tone-shaping tool. (shure.com)


6. Cut first; boost carefully

When a frequency range is too prominent, cutting it usually preserves more headroom than boosting everything around it.

Suppose speech lacks clarity because the system has excessive energy around 300 Hz. Possible approaches:

  • Boost 2–4 kHz by 6 dB.
  • Cut 300 Hz by 3 dB and reassess.

The second approach may produce better clarity without pushing the signal closer to clipping or making sibilance painful.

Boosting also asks the loudspeaker and amplifier to produce more energy. A 6 dB boost requires roughly four times the amplifier power at that frequency for the same input level.

Useful starting discipline:

  • Use the fewest filters necessary.
  • Make modest adjustments.
  • Prefer cuts over large boosts.
  • Bypass the EQ and compare frequently.
  • Measure again after each meaningful change.

If bypassing the EQ makes the system sound more natural, the impressive curve has lost the argument.


7. What EQ cannot fix

Equalization cannot reliably correct:

  • Long reverberation time
  • Poor microphone placement
  • Loudspeakers aimed at microphones
  • Insufficient loudspeaker coverage
  • Severe seat-to-seat level differences
  • Mechanical rattles
  • A deep acoustic cancellation caused by destructive interference
  • A participant speaking three feet from a tabletop microphone

A deep cancellation is particularly dangerous. If two sound paths cancel at one location, boosting that frequency sends more energy into both paths, which may continue cancelling. It also wastes amplifier and loudspeaker headroom.

Shure cautions against relying solely on EQ or feedback reduction when microphone and loudspeaker placement creates an unstable system. (service.shure.com, service.shure.com)

Fix physical problems physically whenever possible.


8. Separate the reasons for EQ

DSP files should distinguish among three different jobs.

Microphone EQ

Follows the microphone channel and corrects:

  • Microphone response
  • Mounting effects
  • Lectern or table resonance
  • Unwanted low-frequency pickup
  • Broad speech tonal problems

This belongs in the microphone input strip, generally after AEC when using a separate AEC block, subject to manufacturer guidance.

Loudspeaker voicing

Corrects and protects a specific loudspeaker system:

  • Driver response
  • Crossover
  • Required high-pass filtering
  • Manufacturer presets
  • Limiting
  • Phase or FIR processing

This belongs at the loudspeaker output and should normally be protected from user adjustment.

Room EQ

Addresses repeatable response problems created by the loudspeaker-room combination.

This belongs after the room or zone mix and before the applicable output. It should be based on measurements from multiple representative positions, not one microphone placed in the room’s acoustically weirdest chair.

Use separate blocks and names:

  • EQ_MIC_LECTERN
  • EQ_ROOM_A_ZONE_1
  • VOICE_CEILING_SPK_MODEL
  • HPF_CEILING_SPK_PROTECT

That separation lets another engineer understand why each filter exists.


9. Practical example: MXA microphone channel

Suppose an MXA ceiling microphone sounds boomy because of HVAC rumble and the room’s reflective ceiling.

A reasonable process is:

  1. Confirm the beam or coverage area targets the talker.
  2. Check raw microphone gain.
  3. Engage an appropriate low-cut or HPF.
  4. Listen and measure again.
  5. Apply a modest parametric cut only if a repeatable resonance remains.
  6. Compare EQ bypassed and enabled.
  7. Test AEC, automixing and quiet talkers afterward.

Shure’s MXA processing provides parametric, low-cut, low-shelf, high-cut and high-shelf filter types. (pubs.shure.com)

Do not immediately carve six deep notches into every lobe. If all lobes show the same severe problem, investigate the room, reference path or output system.


10. Practical example: Q-SYS or Biamp loudspeaker output

Consider:

Room source mix → user volume → room EQ → speaker voicing → limiter → amplifier

The manufacturer’s speaker preset should remain identifiable and protected. Add project-specific room correction in a separate block.

This provides three benefits:

  • Manufacturer voicing can be updated without destroying room tuning.
  • Room EQ can be bypassed independently during troubleshooting.
  • The commissioning report can list only the site-specific filters.

For divisible rooms, keep physical loudspeaker voicing unchanged between modes. Switch room-EQ presets only when measurements demonstrate that opening the wall creates a meaningful, repeatable change. Avoid changing EQ merely because the room-combine Boolean looked lonely.


Practical exercise

Open an offline Q-SYS, Biamp or similar DSP design and create one parametric EQ band at 1 kHz.

  1. Set gain to –6 dB and Q to 0.7. Observe the broad curve.
  2. Change Q to 5. Observe the narrow curve.
  3. Return gain to 0 dB.
  4. Create an HPF at 100 Hz.
  5. Compare 6, 12, 24 and 48 dB/octave slopes.
  6. Add a high shelf at 4 kHz with –3 dB gain.
  7. Bypass each band individually and then bypass the entire EQ.

The objective is to connect the numbers with the response graph before attempting to tune an actual room.

Compact glossary

  • EQ: Equalization; frequency-dependent gain adjustment.
  • HPF: High-pass filter; attenuates low frequencies.
  • LPF: Low-pass filter; attenuates high frequencies.
  • Cutoff: Frequency region where a pass filter begins transitioning.
  • Slope: Rate of attenuation, commonly expressed in dB per octave.
  • Parametric EQ: Filter with adjustable frequency, gain and bandwidth.
  • Graphic EQ: Fixed-frequency, fixed-width filter bank.
  • Q: Measure of how narrowly a filter is focused.
  • Bandwidth: Frequency range affected by a filter.
  • Shelf: Filter affecting everything above or below a chosen frequency.
  • Voicing: Processing intended for a particular loudspeaker system.
  • Notch: Very narrow cut used for a specific resonance or feedback frequency.

Review questions

  1. What happens to bandwidth when Q increases?
  2. Why is a parametric EQ usually better for a narrow resonance than a graphic EQ?
  3. Why should loudspeaker voicing and room EQ use separate DSP blocks?

Answers

  1. The bandwidth becomes narrower, affecting a smaller range of frequencies.

  2. A parametric EQ can target the precise frequency and width of the problem, minimizing changes to adjacent frequencies.

  3. Loudspeaker voicing belongs to the loudspeaker model, while room EQ belongs to the installed acoustic environment. Separate blocks make updates, troubleshooting and documentation clearer.