Single Band vs Multiband Compression for Dialogue
Single-band compression controls a signal as one frequency range. Multiband compression divides it into ranges that can respond separately. For dialogue, start by asking whether the problem affects the whole voice or only a particular part of its sound.
More bands do not automatically produce a better recording. They provide more controls and more decisions to check.
Adobe’s compression reference documents both single-band and multiband processing. This guide offers a listening workflow rather than a universal preset.
Describe the problem without naming a processor
Listen to the unprocessed passage first.
Does a whole phrase jump in level? Does one section of the voice become prominent only on certain words? Is the apparent problem actually a microphone movement or a damaged recording?
Write a concrete observation such as “the lower part of the voice becomes too prominent on this phrase”.
That gives you a testable aim. “Needs multiband compression” chooses a tool before establishing what the tool should accomplish.
Keep the original source available throughout the comparison.
Try the simpler appropriate treatment
If the whole phrase needs a level change, a local adjustment may be enough.
If you want to manage broader changes in the voice’s dynamics, test a modest single-band treatment. Listen to a complete answer rather than one isolated word.
Do not add another processor merely because the first comparison is subtle. Subtle processing may be exactly what the passage needs.
Also check whether the original variation belongs in the delivery. An emphatic word can be part of the speaker’s meaning.
Understand the extra multiband decisions
Multiband processing allows different frequency ranges to have different compression settings.
That can be useful when you have identified a changing problem in a particular range. It also means you need to review the boundaries and behaviour of those ranges.
Do not treat a multiband preset as several independent repairs that require no listening. The combined result still needs to sound like the intended speaker.
Begin with one clear purpose and resist adjusting every band simply because the controls are available.
Choose a representative comparison passage
Use a passage that includes the troublesome words, ordinary speech and a pause.
Keep its position and playback level stable while comparing treatments. If one version is much louder, you may prefer its level rather than its handling of the actual problem.
Make a simple record:
| Version | What to listen for |
|---|---|
| Original | The specific variation you identified |
| Simple treatment | Whether it solves enough of the problem |
| Multiband test | Whether the targeted change helps elsewhere too |
These are comparison stages, not a required processing chain.
An interview example
Imagine a contributor leans closer during one part of an answer. That phrase has a different character from the surrounding speech.
First assess whether another take or a local level adjustment offers a cleaner solution. Do not assume compression can make two microphone positions identical.
If the change remains distracting, compare a restrained single-band treatment with a targeted multiband version.
Then play the preceding and following sentences. A setting that improves the isolated phrase may make the ordinary voice sound unnecessarily altered.
Keep the least complicated version that serves the complete answer.
Listen beyond the selected range
A useful result must work across more than the problem moment.
Review quieter delivery, stronger emphasis and words with a different character. Listen for changes that make the speaker seem less natural or inconsistent from phrase to phrase.
If the treatment only belongs on one passage, consider applying it locally rather than changing the entire interview.
Keep any automation or local processing clearly identified. A later editor should know which section needed attention and why it differs from the rest of the session.
Keep compression separate from fixed tonal adjustment
A persistent tonal imbalance and a changing dynamic event are different observations.
You may need to consider equalisation, a source edit or a different recording rather than another compression stage.
Avoid changing EQ and compression together during the first test. If the result improves, you otherwise cannot tell which change mattered.
Once you understand the separate contributions, review their combined effect. The final sound is what matters, but a controlled comparison makes that judgement easier to explain and revise.
Do not copy settings between speakers blindly
Two contributors may have different deliveries, recordings and roles in the film.
A setting developed for one person is a starting reference at most. Check the next voice from its original recording and identify its actual needs.
For a conversation, compare speakers in sequence. A highly treated voice beside an untouched one may create a distracting difference even when each seems acceptable alone.
The objective is a usable conversation, not identical processing labels across every track.
Reintroduce the soundtrack
Bring music, effects and backgrounds back after the voice treatment is stable.
Listen to the passage at an ordinary playback level. If words remain hard to follow, review the arrangement and balance rather than making the voice progressively more processed.
A different music section may leave more room for the delivery. Choose accompaniment against the actual treated voice, not against a silent picture.
Keep the approved reference so later revisions can be compared with a complete mix.
Check the delivered file
Export a representative section and listen outside the application.
Check the troublesome phrase, ordinary speech and transitions into any locally treated passage. Retain the original source and a clear description of the chosen treatment.
There is no requirement to use multiband compression on every video. Use it when its separate controls solve a specific problem more usefully than the simpler alternatives.
Browse RouteNote Licensing’s categories with the finished narration ready. A well-chosen accompaniment and a deliberate voice treatment should support the same complete scene.