FAIRLIGHT / DYNAMICS / 06.02
Compressor
Manage level. Preserve expression.
From a specific voice-level problem to a controlled comparison. Learn the channel compressor without confusing it with a limiter, a noise remover or a whole-mix acceptance test.
- 01Separate controls and scope
- 02Understand the timing response
- 03Compare before accepting
Threshold and ratio change the static curve, not just the caption.
Threshold / TARGET↗
Separate the static target from the timing response.
Choose one channel.Change the controls to explore the diagram. No audio playback, native DSP or connection to Resolve.
Reference scope · 7 October 2026
Reviewed on 7 October 2026 against the manufacturer overview and relevant primary-manual sections reproduced on iMaster. This mirror is not Blackmagic hosting. Both window illustrations were inspected; exact local build compatibility and listening tests are not certified. Full official PDF retrieval failed. All interactive models are authored simplifications, not native DSP.
Read the source notes ↓On this page
01 / UNDERSTAND & APPLY
A compressor addresses level—not every recording problem
Compression is useful when an edited voice has syllables or phrases that jump out while quieter parts disappear. Check the edit and relative clip levels first. Adjusting one unusually loud sentence with Clip Volume may be more appropriate than asking a compressor to correct the entire conversation.
This guide covers the Compressor inside the channel Dynamics window, not the separate Multiband Compressor insert or a universal voice-processing recipe. Open it by double-clicking Dynamics on the chosen track or bus. Read the channel name in the window before changing anything.
Track
Useful for a sequence of related clips, although one threshold can behave differently with very different recordings.
Bus
Acts on the combined signal of its routed tracks. A loud contributor may affect the whole group, so decide whether shared control is actually intended.
Not restoration
Clipping, noise, room reflections and editing mistakes require their own treatment. A more even level does not by itself mean a clean recording.
Reference: Blackmagic Design — Fairlight / Dynamics · Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror)
02 / UNDERSTAND & APPLY
Reading Dynamics: identify the compressor controls
Expander/Gate, Compressor and Limiter share one window. During an initial comparison, keep the other modules unchanged and alter only the compressor. Toggling the entire Dynamics section at the same time prevents you from identifying which process caused the difference.
The published manual illustration shows a music track with Music pumper* selected. It illustrates the interface, not a default narration preset. The common Make Up control follows Dynamics; it should not be treated as an output control belonging only to the compressor.
| Element | How to read it |
|---|---|
| Input / Output | Levels before and after the relevant processing, not a measurement of the delivered file. |
| Graph | Horizontal: input level. Vertical: output level. Not a frequency spectrum or clip waveform. |
| Gain Reduction · C | Compressor attenuation. Not the total change after Make Up and other modules. |
| Module / overall enable | Bypassing one process and bypassing the whole window are different comparisons. |
| A/B / Reset / Preset | Preserve the initial state before comparing. Reset is not a universal return to your previous custom state. |
Dynamics · Compressor
Dynamics on a music track with Music pumper* selected. Compressor, common Make Up and Sidechain are visible. This is not a default voice preset.
Image source ↗Reference: Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror) · Blackmagic manual illustration — Dynamics / Music pumper (iMaster mirror)
03 / UNDERSTAND & APPLY
Threshold, Ratio, Knee, Mix and Make Up
Threshold marks the transition into the compression characteristic; it is not a guaranteed output ceiling. Ratio describes changes above threshold. For a sustained −12 dB input, a −24 dB threshold and 4:1, an ideal static model gives −21 dB: a 12 dB excess becomes 3 dB. This calculation excludes timing and compensation.
Knee changes the transition around the threshold. Do not confuse it with Attack: the transfer shape and the speed of gain changes are different. Listen to quieter syllables near threshold as well as the loudest peak.
For the documented channel Dynamics, Mix 0 corresponds to compressed sound and Mix 100 to the original signal. Do not import a “100% means fully wet” convention from another plug-in. Check both endpoints in your version before comparing a blend, then restore your chosen value.
Make Up raises the post-Dynamics level. Establish the compression behavior first, then match the perceived level of the compared versions. Do not automatically restore every decibel of attenuation: the goal is intelligible speech with headroom, not identical peak heights.
Reference: Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror)
04 / UNDERSTAND & APPLY
Attack, Hold and Release: preserve the phrase
Attack governs the response when more attenuation is needed. Faster is not always better: compare consonants, impacts and the natural onset of words. A slower response may leave an initial peak; a faster one may change its character. It is not simply a timer that waits and then switches gain instantly.
Hold maintains attenuation before recovery. Release controls recovery. Use a phrase containing a pause: a short recovery can make the background pulse, while an overly long one can leave the next phrase attenuated. Listen to breaths between words rather than relying only on meters.
Milliseconds do not describe the entire detector algorithm. The panel below uses one explicitly defined mathematical response. Matching its numbers to Resolve knobs does not establish identical envelopes in the application.
Reference: Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror)
05 / UNDERSTAND & APPLY
Sidechain: decide what controls attenuation
For ordinary voice leveling, start with the material itself controlling the compressor. An external source serves a separate task, such as speech controlling music attenuation. Process the music and choose the exact voice source, not the reverse.
The illustrated sidechain has On, Listen and Source. On enables the selected control path; Listen helps audition the detector source. Listening to a sidechain is neither a final-mix check nor, by itself, proof of compression. Compare older instructions with your installed build.
Check speech, a pause and a scene transition. Turn Listen off before judging the normal mix. The dedicated Ducker has its own article; avoid combining two control processes until the contribution of each is understood.
Reference: Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror) · Blackmagic Design — Fairlight / Dynamics
06 / UNDERSTAND & APPLY
First pass: one goal, one variable at a time
Prepare material
Save a project version. Choose quiet and loud speech, a breath and a pause. Note whether the problem is an isolated jump or persistent unevenness.
Confirm scope
Check the track or bus in the Dynamics title. Record Gate/Limiter and automation states. Begin without adding compensation gain.
Find useful behavior
Change threshold and ratio separately. Observe which words trigger attenuation. Do not force all speech toward an arbitrary gain-reduction target.
Tune timing
Compare the word onset, a sustained vowel and recovery in the pause. Then listen to the first word of the following sentence.
Compare fairly
Match listening level, compare A/B, then restore the full context with music. “Louder” alone is not evidence of improvement.
Reference: Blackmagic Design — Fairlight / Dynamics · Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror)
07 / UNDERSTAND & APPLY
Teaching envelope: attenuation over time
Enter a quiet level at or below threshold, a louder pulse and its duration. The dashed trace is the instantaneous static target; the solid line is calculated attenuation. Here Attack and Release are exponential time constants in the gain-reduction dB domain; Hold freezes the pulse-end value.
Change one parameter and observe whether attenuation approaches the target before the pulse ends. The graph ends after four Release time constants: the remaining small value is not forced to zero. This is neither a measurement nor a sound-quality model.
ATTACK · HOLD · RELEASE / ENVELOPE LAB
How attenuation builds and recovers
An invented detector-level pulse, not audio. First-order gain reduction in dB; Hold freezes the pulse-end value. These time constants are not calibrated Fairlight knob values.
The bounds above belong to this lesson, not to Resolve. Use decimal dot or comma, without blank values or exponent notation.
No audio files, microphone access, playback or Resolve connection. No automatic saving or transmission. JSON is a teaching model, not a native preset or acceptance evidence. Leaving the page discards edits.
08 / UNDERSTAND & APPLY
Symptom → check → avoid
| Symptom | Check | Do not automatically |
|---|---|---|
| No audible change | Correct channel, enable state, threshold crossings, Mix and sidechain source. | Do not maximize ratio instead of diagnosing. |
| Breaths and background stand out | Compensation, recovery, source noise and clip balance. | Do not add a gate as an obligatory cure. |
| The voice lost expression | Word onsets, attenuation duration and expressive contrast. | Do not equate steady meters with good speech. |
| Output still overloads | Make Up, other processors, the bus and the actual output peak. | Do not treat the compressor threshold as a file ceiling. |
09 / UNDERSTAND & APPLY
Save a decision, then verify the result
Record the purpose, channel, key parameters and the states of the other Dynamics modules. Test preset restoration in a safe copy: a preset name does not reveal everything saved alongside the compressor.
Export a short check through the intended delivery route. Listen to the whole file, especially onsets, pauses and tails. Loudness and peak acceptance is separate from compressor setup; this article does not prescribe one target for every destination.
Reference: Blackmagic Design — Fairlight / Dynamics
PRACTICE / NO AUTOPILOT
A small task. A checkable result.
Work in a test copy, listen before and after, and write down what you actually observed. The checklist contains no pre-approved results.
Define the test
Choose 20–30 seconds containing two differently leveled phrases and a pause. Record the channel, range and one specific problem. Another person should be able to find the same passage.
Separate the static decision
On a copy, choose a sustained vowel. Change only threshold, then only ratio, noting which words trigger attenuation. Greater attenuation alone is not an improvement criterion.
Compare timing
Preserve version A. Change one timing control in B. Note differences at word onset, during the pause and at the next word. Name an audible consequence rather than just “better.”
Check Mix in your version
Record its starting position. Compare 0 and 100 while leaving the other modules unchanged, then restore the chosen value. Distinguish controls you tested from documentation-only notes.
Accept the test file
Export the selected range through the intended routing. Listen from start to finish, noting consonants, breaths and background recovery. Leave the file status unverified until this is done.
FAQ / BEFORE THE NEXT MOVE
Questions that matter in practice
Is compression normalization?
No. Here normalization applies gain based on a measured characteristic, while compression responds to level over time. Neither replaces editing and listening checks.
Does a −12 threshold guarantee nothing exceeds −12?
No. Finite ratio, timing, Mix and Make Up do not provide that guarantee. Verify output peak requirements separately.
Why does Mix 100 remove compression?
The reviewed channel-Dynamics description runs from compressed at 0 to original at 100. This is not a universal Mix-knob convention.
Reference: Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror)
Will a compressor remove noise in pauses?
That is not guaranteed. Compensation may make noise more apparent. Diagnose the source and evaluate specialized treatment separately rather than enabling a whole chain at random.
Is external sidechain always necessary?
No. Define the task first. External control is useful when one source should control another; it is not mandatory for ordinary voice leveling.
Can I transfer the panel curve to Resolve?
No. Its JSON describes an authored exponential teaching model, not a preset or a measured application response. It illustrates attenuation timing.
REFERENCES / 2026-10-07
What this guide is based on
Reviewed on 7 October 2026 against the manufacturer overview and relevant primary-manual sections reproduced on iMaster. This mirror is not Blackmagic hosting. Both window illustrations were inspected; exact local build compatibility and listening tests are not certified. Full official PDF retrieval failed. All interactive models are authored simplifications, not native DSP.
Blackmagic Design — Fairlight / Dynamics ↗
Reviewed 7 October 2026: opening channel Dynamics, module scope, comparison and mixer context. Not an installation test.
Blackmagic — Resolve 21 manual, ch. 177 / Dynamics (iMaster text mirror) ↗
Primary text on third-party hosting, Dynamics section (printed pp. 4017–4022). Checked Mix 0 compressed / 100 original, common Make Up, compressor controls and On/Listen/Source. Not proof of the installed build.
Blackmagic manual illustration — Dynamics / Music pumper (iMaster mirror) ↗
Image inspected: music track, compressor, common Make Up and sidechain. Not a voice preset. Image loads only on request; bytes are not bundled.