Capture and dynamics production workshop / 6 MIN READ
Calculate the compressor curve before adding make-up gain
A reconstructible level-transfer comparison before a sound owner chooses processing.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-07
A familiar control label does not explain which parts of a recording a proposed treatment changes.
What you will leave with: A reconstructible level-transfer comparison before a sound owner chooses processing.
Use this when
- A processing request mixes range control with an overall level adjustment.
- A preset sounds different but its intended operation has not been documented.
Not a time-domain compressor implementation, a listening assessment or a destination loudness certification.
Specify the limited calculation
MathWorks gives the hard-knee steady-state ratio. With threshold T = −20 dB, ratio R = 4 and no make-up, use y = x when x ≤ T; otherwise y = T + (x − T)/R.
Ask the sound owner what problem the request is meant to solve. A conspicuous passage, an uneven performance and a delivery requirement can lead to different decisions. Preserve the untouched recording and the stated purpose before discussing a preset. The calculation here makes one proposed operation inspectable; it does not decide whether that operation is appropriate for the programme.
Use a set of invented input levels rather than a private voice recording for the first comparison. Label the reference and keep it consistent throughout the table. Do not mix quantities from different meters simply because both use the same unit label. The reviewer needs to know what each supplied number represents before interpreting the result.
Build the table rather than guessing from the knob
Inputs −28, −20, −12 and −4 dB give outputs −28, −20, −18 and −16. For −12, the excess is 8; dividing by 4 gives 2, so output is −18. Adding 6 dB afterward gives −22, −14, −12 and −10. This shifts the curve without changing its slopes.
Keep three columns: supplied value, result before the additional adjustment and final result. This lets a reviewer locate an arithmetic disagreement without replaying a recording or guessing which setting was active. Work through one row aloud, then ask a second reviewer to reconstruct a different row. A table copied from a screenshot is less useful than a calculation another person can explain.
The local script evaluates the specified rule and independently checks its branch boundaries and arithmetic. It does not run MATLAB, a plug-in or an audio workstation. No waveform was processed and no playback was assessed. Keep those limits attached to the result so a correct table cannot be mistaken for a successful production mix.
Use the comparison to clarify the request
Ask whether the owner wants the relationships demonstrated in the table. If not, revise the proposal explicitly rather than choosing a different preset until the display looks reassuring. Record which input cases prompted the concern. A quiet-looking meter or a visually compact waveform is not an explanation of why the chosen treatment serves the programme.
Compare a proposed constant adjustment as a separate alternative, with the same supplied values. Do not silently combine several controls and then describe the outcome as a single change. The useful decision concerns what the owner intends to alter and what should remain protected, including the meaning and intelligibility of the recorded contribution.
Leave questions about actual transient behaviour and listening quality to a controlled production review. The static exercise does not establish them. Request a reversible working copy and representative passages if the authorized sound owner proceeds. Keep any processing that changes editorial meaning separate from a technical convenience; unclear speech cannot be made truthful merely by selecting a stronger treatment.
Connect the calculation to the real deliverable
Give the sound owner the proposed operation, the invented table and the unresolved production question. Ask for the actual settings and a comparison appropriate to the source material. Do not instruct them to reproduce the toy numbers on a real voice if the recording, measurement basis or intended treatment differs. The numerical example is a communication aid, not a universal recipe.
Review the resulting export through the team’s established listening and delivery process. Keep the original available and document any accepted tradeoff. A technically explainable operation can still be unsuitable for the story, and a pleasant preview does not establish compliance with a destination requirement. Those decisions require their own evidence rather than a borrowed approval from this worksheet.
Archive the calculation with the reason for the chosen treatment and the exact production version reviewed. When another editor revisits the material, they should understand what the team intended, not merely see an inherited preset name. Close with a bounded statement: the proposed arithmetic is understood, while the actual listening result and release decision remain the responsibility of the appropriate owners.
Make the reasoning concrete.
A fictional narration team asks for a stronger treatment and a louder export in the same sentence. The sound editor needs to know which consequence the producer actually wants.
Increase several controls together
Opening: Choose a more forceful preset and compensate by ear.
Hypothesis: The preview may seem improved quickly.
Risk: The team cannot explain which operation produced the change.
Review the proposed table
Opening: Clarify the intended treatment before processing the recording.
Hypothesis: The owner can discuss a concrete consequence.
Risk: The real performance still needs careful listening and delivery review.
The producer clarifies the request with the sound owner before any processing is applied to the working copy.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Purpose
What programme problem should the treatment address?
Basis
What does each supplied level represent?
Rule
Can the reviewer reconstruct the chosen branch?
Alternative
Which single change is being compared?
Review
What listening and delivery evidence remains necessary?
Watch for the wrong conclusion.
- Treating a preset name as an explanation.
- Combining changes without preserving the comparison.
- Calling static arithmetic a measured listening result.
Your next action
Reconstruct the four rows and ask the sound owner which demonstrated consequence matches the intended request.
Sources & scope
Original fictional production situations and numerical fixtures. Named sources support the specified mechanisms. Local arithmetic is not a hardware, application or listening test. No production settings or assets changed.
- Compressor →Primary documentation · Supports: Supports the static characteristic and make-up adjustment.