Dynamics processing 10 min read
Compressors, Limiters and Automatic Gain Control
Control level changes with threshold, ratio, timing and gain settings while preserving speech clarity and protecting outputs.
- Explain threshold, ratio, attack, release, knee and makeup gain.
- Calculate the effect of a compressor ratio.
- Distinguish compression, limiting and AGC.
- Place dynamics processing correctly in a DSP signal flow.
- Commission dynamics blocks using meters and real program material.
Why an AV engineer should care
Real audio does not remain at one convenient level.
One person speaks quietly. The next person leans into the lectern microphone and attempts to address the entire continental United States. A video contains both whispered dialogue and a thunderous intro. Someone drops a wireless microphone.
Dynamics processors manage these changes:
- A compressor reduces excessive variation.
- A limiter prevents peaks from crossing a defined ceiling.
- Automatic gain control moves signals toward a target level over time.
These tools can improve consistency and protect equipment. Poor settings produce pumping, clipped syllables, elevated room noise and speech that sounds like it has been laminated.
1. Dynamic range
Dynamic range is the difference between a signal’s quiet and loud portions.
Consider a lectern microphone:
- Ambient room noise: –60 dBFS
- Quiet talker: –35 dBFS
- Typical talker: –24 dBFS
- Excited talker: –12 dBFS
- Microphone impact: –2 dBFS
If input gain is raised to help the quiet talker, the excited talker may clip. If it is lowered for the excited talker, the quiet talker becomes difficult to hear.
A compressor reduces the difference between loud and quiet material by attenuating the loud portions once they cross a threshold.
It does not repair analog clipping that already happened at the microphone preamp or ADC. Proper input gain comes first.
2. Threshold and ratio
The threshold is the level where compression begins.
Below the threshold, a standard downward compressor usually passes audio unchanged. Above it, the ratio determines how much the output can continue increasing.
Common ratios include:
| Ratio | Meaning |
|---|---|
| 1:1 | No compression |
| 2:1 | 2 dB above threshold produces 1 dB of output increase |
| 4:1 | 4 dB above threshold produces 1 dB of output increase |
| 10:1 | Strong compression approaching limiting |
| 100:1 or ∞:1 | Essentially limiting |
Suppose:
- Threshold: –20 dBFS
- Ratio: 4:1
- Input: –8 dBFS
The signal is 12 dB above threshold:
12 ÷ 4=3 dB
The output rises only 3 dB above the –20 dBFS threshold:
-20+3=-17 dBFS
The compressor therefore applies 9 dB of gain reduction.
Q-SYS and Biamp define their compressor ratios using this same relationship between input change and output change above threshold. (help.qsys.com, tesira-help.biamp.com)
3. Attack time
Attack controls how quickly the compressor applies gain reduction after the signal crosses the threshold.
Fast attack
A fast attack catches:
- Microphone impacts
- Plosives
- Sudden shouting
- Sharp program peaks
But an excessively fast attack may:
- Remove natural consonant transients
- Make speech dull
- Flatten percussion
- Produce distortion on low-frequency material
Slow attack
A slower attack lets the initial transient pass before compression takes effect.
That may preserve speech articulation and musical punch, but it can allow dangerous peaks through.
The correct attack depends on the purpose:
- Speech consistency: moderate attack
- Loudspeaker protection: fast peak control
- Music: enough attack to preserve desired transients
- Microphone-drop protection: probably faster than whatever the user is planning to do next
4. Release time
Release determines how quickly the compressor stops attenuating after the signal falls below the threshold.
Release too fast
The gain moves rapidly between compressed and uncompressed states, causing:
- Pumping
- Breathing
- Audible level modulation
- Distortion
- Room noise swelling between words
Release too slow
The compressor remains engaged after the loud event has passed:
- The next sentence may be too quiet.
- Music may lose movement.
- A single shout may suppress several following words.
Release should normally sound smooth while recovering quickly enough for the source.
Watch the gain-reduction meter while someone speaks normally. The meter should move deliberately and return without frantic twitching or remaining pinned down forever.
5. Hard knee and soft knee
The knee describes how abruptly compression begins around the threshold.
Hard knee
The selected ratio begins immediately when the threshold is crossed.
Useful for:
- Firm control
- Peak management
- Effects requiring obvious compression
- Situations where transparency is secondary
Soft knee
Compression begins gradually around the threshold and transitions smoothly toward the selected ratio.
Useful for:
- Conferencing speech
- Program audio
- Subtle level control
- Situations where the processing should remain unobtrusive
Biamp describes knee softness as the transition between the uncompressed and compressed portions of its response curve. Q-SYS similarly provides a soft-knee range around the threshold. (help.qsys.com, tesira-help.biamp.com)
6. Makeup gain
Compression reduces loud portions, so the resulting signal may sound quieter overall.
Makeup gain raises the output after compression.
For example:
- Compressor routinely applies 3–5 dB of gain reduction.
- Makeup gain adds 3 dB.
- Average speech becomes more consistent and slightly louder.
- Peaks remain controlled.
Makeup gain can also defeat the purpose of compression.
If a compressor removes 6 dB from peaks and someone adds 10 dB afterward because “more level,” the system now has less headroom than before.
Compare bypassed and processed signals at approximately equal perceived loudness. A louder signal nearly always sounds temporarily better, which makes gain-mismatched comparisons rather dishonest little salesmen.
7. Compressor versus limiter
A compressor manages dynamic range over a useful operating region.
A peak limiter establishes an upper ceiling.
| Compressor | Limiter |
|---|---|
| Moderate ratio | Very high ratio |
| Shapes normal dynamic range | Controls exceptional peaks |
| May use makeup gain | Usually no makeup gain |
| Often placed on speech or program buses | Often placed near system outputs |
| Intended to work periodically | Ideally engages only when needed |
A limiter may use look-ahead, delaying the main signal briefly so it can see an incoming peak before processing it. Biamp’s peak limiter offers a 1 ms look-ahead option for more transparent limiting at the cost of 1 ms additional latency. (tesira-help.biamp.com)
A limiter is not permission to operate a system against the ceiling continuously. If its gain-reduction meter stays active throughout normal operation, fix the upstream gain structure.
8. RMS versus peak detection
A dynamics processor must measure the signal before deciding what to do.
Peak detection
Responds to instantaneous waveform peaks.
Best suited to:
- Clip prevention
- Driver protection
- Sudden transient control
RMS or average detection
Responds more closely to sustained signal energy.
Best suited to:
- Perceived speech level
- Program consistency
- General compression
- AGC behavior
The current Q-SYS compressor uses RMS detection for its primary level calculations, while a peak limiter is designed around peak control. (help.qsys.com, help.qsys.com)
A signal can have a moderate average level while still producing very high peaks. This is why a speech compressor and final peak limiter can serve different purposes in the same signal chain.
9. Automatic gain control is different
Automatic gain control, or AGC, moves the signal toward a target level:
- Quiet signals receive gain.
- Loud signals receive attenuation.
- Maximum boost and cut limits constrain the adjustment.
A compressor normally attenuates only signals above its threshold. AGC can raise quiet signals as well.
AGC is useful when:
- Talker distance varies.
- Many different people use the room.
- A ceiling microphone covers multiple positions.
- The codec needs a reasonably consistent send level.
AGC can also raise:
- HVAC noise
- Room reverberation
- Projector fans
- Distant conversations
- Someone opening a bag of chips with the subtlety of demolition work
Limit maximum boost. If AGC always reaches maximum boost, the microphone gain, placement or target setting is probably wrong.
Shure’s current MXA920 processing places AGC after automixing so it does not change gating decisions. Its guidance also recommends watching the boost/cut meter and correcting the input fader if AGC constantly hits its limits. (shure.com)
10. Linked versus independent processing
Stereo program channels should normally use a linked compressor.
If left and right channels are compressed independently, a loud event on the left may reduce only the left channel. The stereo image can appear to move toward the right.
Linked processing detects the louder channel and applies the same gain reduction to both, preserving their relationship.
Biamp distinguishes:
- Ganged: Same settings, but channels react independently.
- Linked: Loudest channel controls shared gain reduction.
- Side chain: Another signal controls the compressor. (tesira-help.biamp.com)
For separate microphone channels, independent compression may be appropriate. For stereo music, link the channels.
11. Practical example: MXA920 conference send
A practical processed path could be:
MXA coverage areas → AEC → noise reduction → automixer → AGC → gentle compressor → USB transmit
The AGC brings different talkers toward a common average level. The compressor catches unexpectedly loud speech.
Shure recommends performing a sound check and placing its compressor threshold approximately 3–6 dB above normal talker levels so routine speech remains mostly untouched. (shure.com, pubs.shure.com)
Commissioning should test:
- Quiet seated talker
- Normal talker
- Loud talker
- Two simultaneous talkers
- Chair movement and paper noise
- Far-end audio during double-talk
Do not tune only with someone repeatedly saying “test, test” from directly beneath the microphone.
12. Practical example: loudspeaker output protection
A room output chain might be:
Room mix → user volume → room EQ → speaker voicing → peak limiter → amplifier output
The limiter threshold should be based on:
- Loudspeaker rating
- Amplifier voltage gain
- Maximum amplifier output
- DSP output scale
- Downstream input sensitivity
- Required safety margin
The limiter should protect the weakest relevant component while allowing the system to reach its designed level.
Do not choose the threshold by turning up the system until it sounds alarming and then backing it off a smidge. Calculate the electrical limit and verify it with measurements.
Biamp identifies peak limiting before system outputs as a way to prevent DAC clipping and excessive signals from reaching amplifiers and loudspeakers. (tesira-help.biamp.com)
Practical exercise
Create a compressor in an offline DSP design with:
- Threshold: –20 dBFS
- Ratio: 4:1
- Hard knee
- Makeup gain: 0 dB
Predict the steady-state output:
| Input | Expected output |
|---|---|
| –30 dBFS | –30 dBFS |
| –20 dBFS | –20 dBFS |
| –16 dBFS | –19 dBFS |
| –12 dBFS | –18 dBFS |
| –8 dBFS | –17 dBFS |
Then:
- Feed speech or pink noise through it.
- Watch input, output and gain-reduction meters.
- Compare fast and slow attack.
- Compare 50 ms and 1-second release.
- Add 6 dB makeup gain and observe the lost headroom.
- Bypass the compressor while matching apparent loudness.
Compact glossary
- Dynamic range: Difference between quiet and loud signal levels.
- Threshold: Level where dynamics processing begins.
- Ratio: Relationship between input and output change above threshold.
- Attack: Time required to apply gain reduction.
- Release: Time required to remove gain reduction.
- Knee: Shape of the transition into compression.
- Makeup gain: Gain applied after compression.
- Gain reduction: Amount of attenuation currently applied.
- Limiter: Processor preventing peaks from exceeding a ceiling.
- AGC: Automatic gain control; moves signals toward a target.
- RMS: Average-energy measurement used to approximate perceived level.
- Look-ahead: Short delay allowing a processor to anticipate peaks.
- Side chain: Signal used to control dynamics applied to another signal.
Review questions
- With a 4:1 ratio, what happens to an input that rises 8 dB above threshold?
- Why can excessive makeup gain defeat compression?
- Why should stereo program channels normally use linked compression?
Answers
-
The output rises only 2 dB above the threshold, producing 6 dB of gain reduction.
-
It raises the entire processed signal, consuming the headroom the compressor created and potentially causing downstream clipping.
-
Linked compression applies equal gain reduction to both channels, preventing the stereo image from shifting when one side becomes louder.
Discuss a project