How to Add Background Music to a Voiceover Without Drowning the Voice

01Why music buries the voice

We've all heard the ad where the music is so loud you miss the offer, or the tutorial where a drum loop fights the presenter. The music usually isn't the problem. Level, choice and timing are. A music bed is called a bed because it supports the voice; it isn't meant to compete with it.

• Level: the music sits at roughly the voice's loudness, so the ear doesn't know what to follow.
• Choice: vocals or bright instruments in the same frequency range as speech blur the words.
• Timing: the bed stays at one level throughout instead of dipping under speech and lifting in the pauses.
• Tempo: a frantic track under a calm read makes listeners tense; a sleepy one flattens an energetic ad.

You don't need a studio to fix this. Any editor with two tracks will do, and the steps below work in CapCut, Premiere Pro, DaVinci Resolve, Audacity and most others.

02Pick the right track before you touch a fader

Half the mix happens before mixing. Choose the right music and levelling is easy; choose the wrong music and no level will save it. The rules we follow:

1. 1No vocals: sung words compete directly with spoken ones. Use instrumental music under any voice.
2. 2Mood matches message: a sale wants a clear beat, a tutorial wants soft piano or pads, a Ramadan brand film might want oud and light percussion.
3. 3Low on events: avoid tracks with sudden changes or a loud solo landing mid-sentence.
4. 4Longer than the voice: leave two or three seconds before the first word and the same after the last.
5. 5Clear rights: don't lift a song from YouTube or a track you found somewhere without a licence that covers your use.

03Generate an original bed with Vocla Music

Instead of an hour in a stock library, describe the track in Vocla's Music tool and get an original instrumental piece. You set the genre, instruments, mood, tempo and where it'll be used, and the track comes back organised into sections, intro, verse, chorus and outro, with timings you can work with.

• Style controls: include or exclude styles and moods to steer the sound.
• Versions: every change creates a new version, and earlier ones stay playable.
• Section notes: keep cue notes on each section.
• Private sharing: send a client a link with an optional expiry and revoke it any time.

Vocla Music is instrumental; singing from your own lyrics is rolling out gradually and appears in the app once it's enabled for your workspace. Commercial use is included on every paid plan: Starter, Professional and Enterprise. Each music generation costs a fixed number of credits, shown before you confirm.

04Levels: how far below the voice should music sit?

No single number fits every track, because music varies so much in density. These are starting points many editors use; your ears make the final call.

If you publish a podcast, Apple Podcasts recommends an overall loudness of around -16 LKFS, plus or minus 1 dB, with true peak no higher than -1 dBFS. That target is for the finished file, not the music alone, and most modern editors have a loudness meter or normalise option that gets you there.

Source: Apple Podcasts support, “Audio requirements” (podcasters.apple.com), as of October 2026. Video and broadcast platforms publish different targets, so check the one you publish on.

05Ducking, step by step

Ducking means the music dips automatically whenever speech starts and comes back up in the gaps. It's what makes professional ads feel like the music makes room for the voice. The general method in most editors:

1. 1Lay out tracks: voiceover on one track, music on the track below.
2. 2Lower the whole bed first: bring the music down to a quiet level before any ducking.
3. 3Turn on ducking: many editors let you mark the voice as the trigger and the music as the target, or offer an auto-ducking option.
4. 4Smooth the moves: keep the dip and recovery gentle so the music doesn't pump between words.
5. 5Fade in and out: a short fade-in at the top and a longer fade after the last word.
6. 6Listen end to end: once on headphones, once on a phone.

No auto-ducking in your editor? Do it by hand with volume keyframes: one point that dips the music just before a line, another that brings it back after. It takes longer, but it's precise.

06In Vocla Video the music ducks itself

If you're turning a script into a video in the Video studio, there's nothing to mix by hand. Switch on a music bed when you set up the project and Vocla lays it under the voice, ducks it under speech in the final mix and burns in the Arabic captions. A music bed costs 120 credits once per video, on top of the video seconds and the voiceover.

A short line like that over a light oud bed is all a 15-second Reel needs. Video starts on Starter; on their own, you can generate the voice on any plan and the music on any paid plan, then mix them yourself.

07When to skip the music

Not everything needs a bed. Sometimes the best call is to let the voice stand alone, and listeners appreciate it:

• Religious content: many viewers prefer religious content without music, out of conviction. A clear, calm voice on its own, or light natural ambience, is a respectful choice.
• Condolences and official notices: sensitive or formal messages need gravity, and music can make them feel promotional.
• Dense explanations: when information comes thick and fast, music distracts more than it helps.
• IVR menus: phone menus are usually clearer without music; save it for on-hold messages.

On-hold audio has its own rules, because phone lines compress the signal and cut frequencies. We cover that in a separate guide.

08Common mistakes and a pre-publish checklist

• Vocals under speech: the fastest way to lose the message. Switch to an instrumental.
• A hard stop: the music cuts with the last word. Give it a fade after the voice ends.
• One level throughout: without ducking, the bed fights the voice.
• Export loudness: the final file is so loud it distorts, or so quiet people max out their volume.
• Unclear rights: an unlicensed track can get a video muted, blocked or claimed.

Before you publish, ask three questions. Did I catch every word on a phone speaker? Does the music fit the mood of the message? Do I have the right to use this music here? Three yeses and you're ready.