For dialogue, a useful starting point is corrective EQ before compression and any final tonal adjustment afterwards. That lets you address a specific tonal problem before setting the compressor, then judge the resulting voice in the mix.

It is a starting point, not a rule. If the recording already sounds balanced, you may not need corrective EQ at all. The best order is the one that solves an audible problem while keeping the speaker natural.

Understand what changes when you swap the order

An EQ changes the balance of frequencies. A compressor responds to its input according to its controls and detector design. Putting an EQ before it changes the signal reaching that input; putting the EQ afterwards changes the compressed result.

iZotope’s guide to EQ and compressor order explains how unwanted low-frequency thumps can influence compression and why upstream EQ can change the response.

That is the important distinction. You are not simply choosing which effect sounds nicer first. You are deciding what signal the compressor receives.

The details depend on the processor. Check its manual if it uses a separate sidechain, automatic gain or another mode that changes the comparison.

Start with one specific complaint

Describe the problem before rearranging the effects.

“The voice sounds uneven” might call for editing levels or adjusting compression. “A dull thump makes the next word dip” suggests a different investigation. “The recording feels too bright throughout” is another problem again.

Choose one representative passage and write a sentence about it.

For example: “At the start of the answer, a low thump is followed by a noticeable change in the voice level.” That gives you something to test. It also stops the session turning into a general search for a more impressive sound.

Keep the untreated recording available. If the problem already exists there, changing the effect order may only be part of the solution.

Try corrective EQ before compression

For the first test, place the corrective EQ ahead of the compressor. Make only the adjustment needed for the identified issue.

Do not add brightness, warmth and extra bass at the same time. You would make it harder to tell which change affected the result.

Listen to the full phrase and watch the compressor’s gain-reduction display. Does the unwanted event still coincide with an obvious movement? Does the speech remain convincing before and after it?

Treat these as observations rather than a demand for a perfectly still meter. A moving display does not automatically mean there is a problem.

If the thump is isolated, compare a local repair with the EQ approach. A permanent tonal change across the whole interview may be unnecessary for a single event.

Try tonal EQ after compression

Now consider the overall character of the processed voice.

Perhaps the level is controlled, but you would prefer a little less weight or a softer top end. An EQ after the compressor lets you judge that tonal adjustment on the compressed result.

For this test, leave the upstream corrective settings alone. Add the smallest useful downstream adjustment and compare it with bypass.

The purpose is not to build a long chain. Two EQ stages are only useful when they have separate jobs. If one small adjustment gives you the result you want, keep the simpler setup.

Recheck the final output level after changing the tone. A comparison is less useful if one version is much louder.

Compare the two orders fairly

Make two versions of the same short dialogue passage.

Version A uses EQ followed by compression. Version B uses compression followed by EQ. Begin with the same settings so you can hear the consequence of the swap.

Then adjust the final playback levels to make the comparison reasonably even. Listen to complete sentences, including pauses and word endings.

Use a short note sheet:

CheckWhat to record
Speech clarityA word or phrase that became easier or harder to follow
Level movementA specific dip, surge or change in emphasis
ToneA repeatable difference rather than “more professional”
BackgroundA pause where room sound became distracting
PreferenceThe version you would keep and the reason

After that initial comparison, optimise the preferred order. Identical settings are useful for demonstrating a change; they are not necessarily the best final settings for both chains.

Keep repair and character separate

A presenter recorded across several days may need different local corrections even when the overall programme should sound consistent.

Avoid solving every clip with one dramatic master-chain adjustment. Select the troublesome passage, identify its cause and make a focused change before judging the shared chain.

Similarly, do not use compression simply because it appears in a tutorial. A well-controlled voice recording may need only small level edits.

For a longer interview, review a quiet answer, an animated answer and the transition between them. An order that works on one sentence may be less suitable for another.

Save the settings with a plain explanation such as “corrective EQ before compressor to reduce the effect of the opening thumps”. The reason is more useful to a collaborator than a preset called “final voice”.

Recheck with music and sound effects

Return the dialogue to the full video. A soloed voice can sound attractive while still sitting poorly against the soundtrack.

Keep the music unchanged for the first comparison. Check whether you can follow the same words with each dialogue version, then decide whether the soundtrack itself needs a simpler arrangement.

Do not keep changing the voice merely to accommodate a busy music candidate. Try a less prominent track as another option.

RouteNote Licensing’s music categories provide starting points for a shortlist. Choose against the edited scene rather than assuming a genre guarantees space for speech.

Know when to stop

The useful result is a clear voice with deliberate, repeatable processing. It does not require a universal effect order.

If you are still unsure, keep the simpler chain and ask a second listener to compare the same passage without seeing which version is playing.

For the controls themselves, read our guides to parametric versus graphic EQ and single-band versus multiband compression. Then return to the recording and make the smallest change that addresses the problem you can actually hear.