How to Mix Audio in CapCut: Voice, Music and Ducking

Use separate tracks before touching volume

Put narration, music and important sound effects on separate timeline layers. On a long project, name or position them the same way throughout: voice at the top of the audio stack, music below it, then effects. That layout matters. Without it, a volume move meant for the music can land on a sound effect or the original camera audio.

If the music isn't in the project yet, start with our separate guide to adding music in CapCut. The steps before and after the mix are in our complete CapCut tutorial. This page picks up once your clips are already on the timeline.

Before mixing:

1. Trim obvious empty space and unwanted starts or endings.
2. Split the music where a section needs its own treatment.
3. Mute duplicate audio tracks that play the same sound twice.
4. Check whether the voice itself clips, distorts or contains strong room echo.
5. Leave the original recordings available so a destructive edit is reversible.

A loudness tool can't recreate a voice that clipped during recording. Noise reduction may help with steady background noise, but heavy-handed processing can damage consonants and leave speech sounding watery. If the recording needs cleanup before the mix, use our CapCut noise-reduction guide.

Set the voice first, then bring in music

Solo the narration, or mute the music for now. Pick a representative section with ordinary speech, a quieter sentence and the loudest word, and listen through it. Set the voice so it's easy to understand without sounding strained or distorted. If several voice clips jump in level, fix those differences before you decide where the music belongs.

CapCut offers loudness normalization to make levels more consistent. It can smooth out distracting jumps between similar voice clips, but it doesn't decide how narration and music sit together. A normalized music track can still mask speech, and a normalized voice can still end up too quiet in a busy mix.

Now unmute the music and start it low. Bring it up until it adds energy, then pull it back if you hear any of these problems:

• words become harder to understand;
• you need to concentrate on the narration;
• consonants disappear beneath drums or vocals;
• the music feels like the subject rather than support;
• the mix sounds acceptable on headphones but speech vanishes on a phone speaker.

There's no useful universal "voice at 100, music at 10" rule. Music with vocals or a dense midrange often fights speech harder than a sparse instrumental at the same displayed level. Judge the passage you're hearing, not the slider number.

Duck the music only where speech needs room

Ducking means lowering the music or ambience during speech and bringing it back up between spoken sections. You're not trying to make the background disappear. It should stay part of the scene and still leave room for every important word.

CapCut's current ducking guide documents both manual volume changes and a mobile Auto Ducking option in supported interfaces. When an automatic control shows up, treat it as a first pass and listen to every transition. If you don't see it, or it lowers the wrong sections, keyframes give you direct control.

Build a four-point ducking envelope with keyframes

CapCut documents keyframes as controls for changing properties over time, including volume in supported audio workflows. On the music track, build each duck around a spoken section:

1. Before speech: place a keyframe at the normal music level shortly before the first word.
2. At the start: place another keyframe after it and lower the music until the voice becomes easy to follow.
3. Near the end: hold that lower level through the final word with a third keyframe.
4. After speech: place a fourth keyframe and return the music gradually to its normal level.

The horizontal gap between each pair sets how fast the level changes. Put the keyframes almost on top of each other and you get a sharp move; spread them out and the ramp gets gentler. Watch breaths and short pauses. If the speaker comes right back in, leaving the music down usually sounds calmer than letting it pump up and down between phrases.

When several short lines sit close together, one longer duck may sound more natural than a separate dip for every sentence. In a long music-only section, let the track come back to its baseline so the edit gets its energy back. Decide by listening. You don't have to draw the same envelope everywhere.

When Auto Ducking helps—and when it does not

Use Auto Ducking if it's there and the result follows the spoken sections accurately. Then check:

• Does the music fall before the first important word?
• Does it stay down through quiet syllables and pauses?
• Does it rise after the thought ends rather than after every breath?
• Does it react to the intended voice track rather than unrelated sound?

If the automatic result pumps, cuts musical accents or misses a quiet line, replace that section with manual keyframes. Availability and naming can differ, so don't waste time hunting for a button you saw in a tutorial from another platform. You can build the same result directly on the timeline.

Use fades for clip edges, not as a substitute for ducking

A fade shapes the beginning or end of a clip. Ducking changes volume around speech inside a longer clip. They solve related but different problems.

CapCut's audio editing guidance describes volume, fade-in and fade-out controls for a selected audio track. Use a fade-in when the music should ease in at the start of a scene, and a fade-out when it should leave without an abrupt stop. Where two music clips meet, overlap or shape their edges only if the transition benefits from it. Not every cut needs a crossfade.

When someone speaks in the middle of a song, keyframes usually preserve the music better than chopping the track up and fading every fragment. Repeated clip-edge fades can leave gaps, uneven timing and more pieces to manage.

A clean sequence often looks like this:

• short fade-in at the start of the music;
• stable baseline under nonverbal footage;
• keyframed duck during narration;
• return to baseline after the line;
• fade-out at the end of the scene.

Improve voice clarity without over-processing it

When the voice is clean but uneven from clip to clip, normalization can help even it out. If it lacks presence, CapCut's Voice Enhancer guidance describes processing meant to improve clarity. Keep the unprocessed version around and compare a short section before you apply the same treatment to the whole project.

Use each tool for the problem it addresses:

If the project is silent, not just unbalanced, work through the CapCut no-sound checklist before you rebuild the mix.

Mix sound effects around the story

A sound effect should be audible for a reason: a transition, action, notification or impact. Place each effect on its own layer when you need independent control. Check it once with the music muted and again in the full mix. An effect that sounds fine alone can turn harsh when it lands on a snare, a loud word or another effect.

Prioritize in this order:

1. dialogue and essential information;
2. effects that explain an action;
3. ambience that establishes the scene;
4. decorative effects and music accents.

That order is editorial guidance, not a CapCut rule. In a music-led montage the music is the main content, so you may flip the emphasis. What matters is which sound the viewer has to notice at that exact moment.

Troubleshoot the mix by symptom

If you only need the finished audio track, the audio-only export guide covers the available routes and format limits. For an ordinary video, check the audio in the exported video itself, since the timing of picture and sound can change how a transition comes across.

Check the saved file before calling the mix finished

Listen to the local export outside CapCut. Export is the final step in CapCut's general editing workflow, but the saved file still deserves its own review.

Use two short passes:

1. Headphones: listen for abrupt ramps, clicks, damaged consonants, duplicate audio and music that changes level for no reason.
2. Phone speaker: check whether every important word remains understandable and whether key effects still register.

Then watch the start and end of each ducked section. The music should make room for speech without sounding as if someone switched it off. If a problem shows up only in the saved file, look at the export or playback stage. Don't start changing a timeline mix that was already balanced.

I'd finish one representative minute before working on the whole project. Set the voice, choose the music baseline, build one good duck, check it on a phone, and carry those decisions forward instead of copying identical numbers. CapCut can handle straightforward narration, music and effects mixes. When you need detailed EQ, compression, routing, metering or repair that your current CapCut interface doesn't expose, move to a dedicated audio editor.