How Loud Should Music Be Under a Voiceover on Reels and TikTok?

The most common audio mistake in short video is not a bad song. It is a good song played two notches too loud. The viewer does not consciously notice the music. They notice that they had to work to understand you, and they scroll.

Here is the short answer. The background music volume under a voiceover should sit low enough that every word is effortless to understand on a phone speaker, which in practice means the music lives well below the voice while you talk and only rises in the gaps where nobody is speaking. If you have to choose between vibe and clarity, choose clarity every time. A viewer forgives quiet music. They do not forgive a sentence they missed.

The rest of this post is about how to get there by ear, why the right level changes depending on what the video is for, and the few moments where the track should actually take over.

Why voice clarity wins the trade

In a voiceover video, the words are the content. The music is the room the words sit in. When the room gets louder than the person talking, the viewer stops following the argument, and a viewer who loses the thread in second four rarely finds it again in second nine.

This matters more on short video than anywhere else for three reasons:

• Phone speakers are small. They compress everything into a narrow band, and that band is exactly where both voices and most music live. On a laptop the voice might separate from the track. On a phone held at arm’s length, it mostly does not.
• Viewers are not in quiet rooms. They are on a bus, in a kitchen, next to a TV. Your mix has to survive background noise you will never hear.
• There is no rewind habit. On long-form video, people scrub back to catch a missed line. In a feed, they swipe.

When we look at voiceover-led videos in our library of analyzed videos, a muddy mix shows up again and again as a reason a strong script still sagged. The hook was fine on paper. It just did not land in the ear. We make a related point about the opening line in visual vs verbal hooks: a verbal hook only works if it is heard on the first try.

A practical rule of thumb for levels

You do not need a meter to mix well, but it helps to know roughly where you are aiming. Most editors who mix voice and music for a living land in a similar place:

• Voice is the loudest thing in the mix, and it stays at a steady level from start to finish.
• Music under speech sits far below the voice, quiet enough that you could describe it as “there” rather than “playing.”
• Music in the gaps (the intro beat, a pause, the outro) comes up noticeably, closer to the voice level, then drops again when you speak.

If your editing app shows volume as a percentage, a useful starting point is to put the voice near full and the music somewhere around a fifth to a quarter of that while you are talking. Then adjust by ear using the test below. The exact number matters less than the gap between the two.

The phone speaker test

Before you post, play the video on your phone, through the speaker, at about half volume, held at arm’s length. Ask two questions:

1. Could I write down every word? If you had to strain for even one, the music is too loud.
2. Can I tell there is music at all? If not, and the track is meant to carry mood, nudge it up slightly.

Then do it once more with a fan or TV running in the background. That second pass catches most mixes that sound fine in headphones and fall apart in the real world. Headphones are the worst place to judge a short video mix, because they flatter it.

Ducking: let the music breathe around the words

The simplest way to get both clarity and energy is ducking. Ducking means the music automatically dips whenever the voice is present and comes back up in the gaps.

Most editing apps let you do this one of two ways:

• Automatic ducking. Some apps can detect speech and lower the music under it. Turn it on, then check that the dip is deep enough and that the recovery is not so fast it pumps audibly between words.
• Manual keyframes. Set a low level for the music during each spoken passage and a higher level for each gap. It takes longer, but you control exactly where the track swells.

A few details make ducking sound natural rather than mechanical:

• Do not duck between every word. Short breaths should stay ducked. Only bring the music up in real pauses, roughly half a second or longer.
• Ease the transitions. A quick fade in and out sounds deliberate. An instant jump sounds like an error.
• Use the gaps on purpose. If you have a line that needs to land, leave a short silence after it and let the music fill that beat. The swell becomes punctuation.

Sound-on vs sound-off: the level depends on who is watching

Not every viewer hears your mix. A share of people watch with the sound off, and the split depends on the platform, the time of day, and where they are. That changes what the music is actually for.

If your video only makes sense with sound, the voice has to carry it, and the music should never compete. If your video makes sense with sound off, through on-screen text and visuals, the music can play a bigger role for the people who do have sound on, because they are not relying on the voice alone.

Our guide on whether captions increase views covers the sound-off side in depth. The short version for this post: captions protect you from a slightly busy mix, but they do not excuse one. A viewer reading along while fighting the music still feels the friction.

When the track should carry the cut

There are formats where music is not background at all. It is the structure. In those videos, the voice steps back and the track takes the lead.

• Montage and transformation videos. Before and after, a room makeover, a week of training. The edit is cut to the beat and the music provides the emotional arc. A voiceover, if there is one, is short and sparse.
• Mood and aesthetic content. Travel, cooking ambience, slow product shots. The track sets the feeling, and words would interrupt it.
• Trend-led videos. If the sound itself is the reason the video gets discovered, it needs to be recognizable, which means audible.

In these cases, flip the rule: the music sits near the top of the mix, and any voice that appears comes in over a deliberate dip. We cover how to cut visuals to the music in editing to the beat, and the choice between a trending sound and your own audio in original audio vs trending sound.

The mistake to avoid is a hybrid where neither leads. A dense script read over a full-volume track gives the viewer two things to follow and no clear priority. Pick one to carry the video.

Choosing music that does not fight the voice

Level is half of it. The other half is what the track sounds like. Some music sits under a voice easily. Some fights it at any volume.

Tracks that tend to sit well under speech:

• Instrumental, no vocals. A singer under a speaker is two voices competing for the same attention. This is the single biggest factor.
• Steady rhythm, simple melody. A track that loops calmly gives the voice a floor to stand on.
• Low and high, not middle. Bass and light percussion leave room for speech. Busy mid-range instruments like piano chords and guitars crowd it.

Tracks that tend to fight the voice:

• Songs with lyrics, especially lyrics in the same language you are speaking.
• Big builds and drops that land in the middle of a sentence.
• Tracks with a strong lead melody that the ear wants to follow.

If you love a track with vocals, use it in the gaps only: the opening beat before you speak, or the outro after your last line. That way the song does its job and never talks over you.

Matching energy to the script

Music volume also changes how the voiceover itself is perceived. A track that is too energetic under a calm, explanatory script makes the speaker sound flat. A track that is too slow under an urgent script makes the speaker sound rushed.

The fix is to match the music’s energy to the delivery, then set the level so the energy is felt but not heard. If you are recording your own voice versus using a generated one, delivery and music interact even more. We compared the two in a TTS voice vs your own, and the choice between narrating and speaking on camera in voiceover vs talking to camera.

A quick checklist before you post

• Voice is steady and on top from the first word to the last.
• Music under speech is quiet enough that you could miss it if you were not listening for it.
• Music rises in the real gaps and falls before the next line starts.
• No lyrics under talking, unless the song is the point of the video.
• The phone speaker test passes at half volume, at arm’s length, with background noise.
• One element leads. Either the voice carries the video or the track does.

How Blossom helps with audio

When you run a video through Blossom, the analysis maps the audio across the timeline, including where the energy builds and drops, with timestamps you can check against your own edit. That makes it easier to see whether a swell in the music landed on a pause or right on top of your most important line, and whether the moments viewers leave line up with a stretch where the mix gets crowded.

It will not tell you a magic volume number, because there is not one. What it gives you is a second pair of ears on where the audio supports the video and where it gets in the way. You can see what the analysis covers on the FAQ.

The short version

Keep the voice on top and steady. Keep the music low enough under speech that nobody has to strain. Let the track rise only in the gaps, and let it lead only in formats built around it. Then test on a phone speaker in a noisy room, because that is where your viewers actually are.