A compressor controls changes in level across a recording. A limiter concentrates on keeping peaks under control, usually with a stronger response. Use compression when speech or other material needs steadier dynamics; use limiting when the mix needs a defined peak boundary.

Neither replaces a sensible balance between dialogue, music and effects.

Start with the problem you can hear

Before inserting a processor, describe the fault.

If one sentence is much quieter than the next, clip gain or automation may be the cleanest first step. If a speaker repeatedly moves between soft and loud delivery, compression may help. If a few impacts create excessive peaks, a limiter may catch them—or you may simply need to turn those effects down.

Do not apply both because a preset chain contains both.

Audacity’s Compressor documentation describes dynamic-range reduction and its threshold, ratio and timing controls. Its separate Limiter guide explains stronger peak control. Their settings and behaviour are specific to those tools, so check the documentation for the processor you use.

Level obvious differences first

Imagine an interview with three sections:

• the speaker begins quietly;
• becomes animated in the middle;
• finishes with a soft final sentence.

Raise the quiet clips and lower the loudest section until the overall performance is closer to the intended balance. Use smooth changes at edit boundaries.

Now the compressor can manage smaller variations. It does not need to react dramatically to an entire section recorded at a different level.

This also prevents room noise from becoming unexpectedly loud when you add make-up gain.

Understand the compressor controls

Threshold determines when processing starts. Ratio describes how strongly level above that point is reduced. Attack and release influence how quickly the change begins and recovers.

A softer knee can make the transition into compression less abrupt. Make-up gain raises the processed output.

These controls interact. A low threshold with a gentle ratio can behave very differently from a high threshold with a strong ratio. Fast timing can change consonants and transients; slow recovery can hold the voice down after a loud word.

Listen to complete phrases, not only a single peak.

Aim for steadiness without flattening expression

A narrator should still be able to stress a word and soften a sentence. Heavy compression can make every syllable equally insistent.

For an ordinary voiceover, begin conservatively. Reduce the stronger moments enough that the listener does not need to keep changing volume. Then compare at a matched playback level.

If the processed version only seems better because it is louder, lower its output and listen again.

This is a practical judgement rather than a universal ratio or gain-reduction target. A quiet documentary interview and an energetic gaming presenter need different treatment.

What the limiter adds

A limiter can restrain brief peaks near a chosen output boundary. In a video mix, those peaks might come from an emphatic word, a musical accent or a layered impact.

Use the delivery specification to choose appropriate peak and loudness settings. A sample-peak ceiling and a true-peak limit are not automatically the same measurement. If the brief specifies true peak, use a tool that measures and controls it appropriately.

Avoid setting a loudness target merely because a social-media tutorial recommends one number for every platform and format.

Do you need both

Sometimes. A voice compressor can smooth delivery while a final limiter catches occasional peaks in the full mix.

But consider what is feeding the limiter. If music, dialogue and effects are all too loud together, lowering the overall bus may be preferable to forcing constant reduction.

If every impact makes the narrator dip or the music audibly breathe, fix the balance upstream. The master processor is revealing a mix problem.

For already mastered library music, start with level changes. Additional compression may remove the contrast that attracted you to the track.

A practical interview example

First, edit mistakes and even out the three sections with clip gain. Next, apply light compression while checking normal speech, emphatic words and pauses.

Add music at a level that supports the interview. Introduce only the effects the story needs. Then inspect the final output for peaks.

If a logo impact is the main offender, turn down the impact. If occasional residual peaks remain, use the limiter gently and recheck the delivery measurements.

Finally, compare the full mix with all processing bypassed at a similar listening level. Confirm that clarity improved and the speaker still sounds human.

Check the export

Render the actual delivery file, reimport it and listen from beginning to end. Encoding, routing or export settings may change the result.

Check the first word, loudest sequence, transitions and final tail. Measure the exported file against the brief rather than assuming the timeline setting guarantees compliance.

Compressor or limiter checklist

Symptom	First action to try
One clip is quieter than the others	Clip gain or automation
Speech varies repeatedly within phrases	Gentle compression
One effect creates a large peak	Lower or reshape that effect
Occasional final peaks exceed the brief	Appropriate limiting
The whole mix feels crushed	Reduce upstream levels and processing
Music masks the narrator	Rebalance or choose a simpler track

Build a clear mix with RouteNote music and SFX before asking processors to finish it. Dynamics tools work best when they solve a defined problem.