If you have ever exported a video, played it back, and wondered why the music suddenly sounds too loud, too quiet, or just plain messy, you are not alone. The question of how loud should background music be under a voiceover sounds simple, but the answer gets confusing fast because people are usually talking about different measurements, different platforms, and different styles of narration.
The good news is that you do not need a perfect studio to get it right. You just need a repeatable way to balance the voice, the music, and the way the platform will compress your final upload.
Why the Numbers You Find Contradict Each Other
Search for guidance on background music levels and you will see advice like minus 18 dB, minus 12 dB, or “keep it low enough that people can still hear the voice.” Those answers are not always wrong, but they often describe different parts of the mix.
That is why the same advice can feel useless in practice. A music bed that works under a calm tutorial voice may completely fail under a fast, energetic voiceover. A soft piano track may sit beautifully under dialogue, while a busy drum loop may fight the voice even at a lower volume.
The real issue is that there is no single magic number that works for every project. Instead, you need to understand the three measurements people mix together when they answer this question.
Three Different Measurements, One Number
When people ask how loud should background music be under a voiceover, they usually mean one of three things:
- The steady level of the music while the voice is talking
- How much the music drops when the voice starts
- How much processing the music has in relation to the voice after mastering or normalisation
If those sound similar, that is exactly why the advice online gets muddy.
1. The steady level under the voice
This is the base level of the music bed while narration is present. The music should support the voice, not compete with it.
2. The duck amount
This is how much the music turns down when the voice begins. Ducking can help the narration feel more present without forcing you to lower the entire track too much.
3. The final loudness after export
Platforms like TikTok, YouTube Shorts, and Instagram Reels may change the way your audio feels after upload. A mix that sounds balanced in your editor can feel different once it is compressed and normalised.
Once you separate these three ideas, the question becomes much easier to answer.
The Gap: How Far Below the Voice
The most practical starting point is this, set the voice clearly first, then place the music noticeably below it.
For most short-form content, a useful approach is:
- Keep the music 10 to 20 dB below the voice on average
- Use the lower end of that range for busy tracks, fast speech, or dense sound design
- Use the higher end for calmer tracks or very sparse arrangements
That range is not a law, it is a starting point. What matters most is intelligibility. If a viewer has to strain to understand the words, the music is too loud, even if your meters look fine.
A quick test is to listen on your phone speaker at a normal volume. If the voice remains clear without effort, you are close. If you can hear the music more strongly than the words, reduce it.
If you are building a content workflow with tools like aicut, this is one of the biggest quality upgrades you can make. Clean voice balance instantly makes AI-generated shorts, UGC-style ads, and social clips feel more professional.
The Duck: How Far It Drops When You Speak
Ducking is the automated dip in music level when narration starts. This is often where editors overdo it or underdo it.
A practical ducking range is usually around 3 to 8 dB depending on the track and the voice. More than that can make the music sound like it is breathing unnaturally. Less than that may not create enough space for the voice.
A simple ducking rule
Use more ducking when:
- The voice is fast
- The music has lyrics or strong midrange instruments
- The video needs a clear instructional tone
- You want the voice to feel dominant and authoritative
Use less ducking when:
- The content is emotional or cinematic
- The music is ambient and minimal
- You want the background to stay present between phrases
- The narration is slower and more expressive
Ducking works best when it is subtle. The audience should feel the music support the voice, not notice the volume automation itself.
Why Over-Ducking Sounds Worse Than No Ducking
Many editors assume more ducking always means cleaner audio. In reality, too much ducking can create a strange, distracting mix.
Here is what over-ducking can do:
- Make the music feel like it is being punched on and off
- Break the emotional flow of the track
- Create an obvious pumping effect
- Make the mix feel artificial, even if the voice is clear
Sometimes no ducking at all is better than aggressive ducking. If the music is already sparse and the narration is slow, heavy automation can make the whole video feel overproduced.
The goal is not to eliminate the music whenever the voice appears. The goal is to make the voice easy to follow while keeping the background energy intact.
Attack and Release: The Settings Nobody Explains
If ducking sounds awkward, the problem is often not the amount of reduction. It is the timing.
Two settings matter most:
Attack
Attack controls how quickly the music drops when the voice starts.
- Fast attack makes the music step out of the way immediately
- Slow attack makes the drop feel smoother, but may let the voice get masked at the start of a sentence
A middle-ground attack usually works best for short-form video.
Release
Release controls how quickly the music rises after the voice stops.
- Fast release can feel jumpy
- Slow release can feel polished and natural
If your voiceover has frequent pauses, a slightly slower release often sounds better because the music returns smoothly between phrases.
A practical starting point
If your editor allows it, start with:
- Moderate attack for quick clarity
- Moderate-to-slow release for smoother transitions
Then listen for whether the music feels like it is “breathing” too obviously. If it does, adjust the attack and release before changing the overall level again.
Set the Voice First, Then the Music
One of the most reliable ways to answer how loud should background music be under a voiceover is to stop thinking about the music first.
Start with the voice.
Step-by-step mix order
- Normalize or level the voice so it is consistent
- Make sure every word is easy to understand
- Add the music at a low level
- Raise the music until the video feels alive, but the words stay clear
- Apply ducking only if needed
- Check the full mix on speakers, headphones, and a phone
This order works because the voice is the message. The music is there to guide energy, pace, and emotion.
If you are producing content at scale, a platform like aicut can make this process much faster because it supports AI video generation, viral prompt cloning, campaign automation, and direct social publishing. That matters when you are testing multiple hooks or ad variations and need the audio to stay consistent across versions.
What Changes When There Is No Narration
Not every short video uses a voiceover. Some rely on text overlays, ambient sound, or visual storytelling alone. In that case, the music can carry much more of the emotional weight.
When there is no narration:
- The music can sit much louder
- You can use more dynamic arrangement changes
- You can allow stronger builds and drops
- You do not need to carve out as much space in the midrange
That said, if the video includes text on screen, the music still should not be so busy that it makes the visuals feel chaotic. For instructional or product videos, calmer tracks usually perform better because they let the viewer process the message.
Loudness Normalisation Moves the Whole Mix
A lot of creators mix audio, export, upload, and then wonder why the balance changed.
That happens because platforms often normalise loudness. In plain language, they may turn the whole video up or down to match their playback standards.
This means:
- A mix that is too quiet may get boosted, but not always in a flattering way
- A mix that is too loud may get turned down
- The relationship between voice and music can feel slightly different after upload
The fix is to leave a little headroom and avoid pushing every element to the max. A balanced mix usually survives platform processing better than an overcompressed one.
Why Turning Everything Up Does Not Help
When a mix feels weak, the instinct is often to make everything louder. That usually creates more problems than it solves.
If you raise the music and the voice together:
- The relationship between them does not improve
- You may increase distortion or harshness
- You can make the voice fatiguing to listen to
- The upload may get normalised back down anyway
Instead of turning everything up, focus on contrast. The voice should feel clearly foregrounded, and the music should feel supportive rather than equal.
This is especially important for TikTok, Reels, and Shorts, where viewers often watch in noisy environments. A clear voice wins more often than a “fuller” sounding mix.
Tracks That Fight the Voice No Matter the Level
Sometimes the issue is not the volume. It is the track itself.
Certain music will clash with narration even when it is quite low.
Common troublemakers
- Tracks with strong vocal chops or lyrics
- Bright synths with lots of energy in the same range as speech
- Busy percussion that keeps drawing attention
- Dense midrange arrangements with little space
If a track fights the voice, lowering it further may not solve the problem. You may need a different piece of music altogether.
Better choices for voiceover content
- Ambient beds
- Minimal percussion loops
- Light rhythmic tracks with a clear low-mid pocket
- Music with gradual movement instead of constant variation
This is why prompt-based workflow tools and quick content testing matter. With aicut, creators can generate and iterate on short-form content quickly, then compare which soundtrack and pacing combination holds attention best.
Checking the Mix on a Phone Speaker
If you only check audio on studio headphones, you are missing the device most of your audience will use.
A phone speaker is a great reality check because it reveals whether the voice is still intelligible when the playback is tiny and imperfect.
What to listen for on a phone
- Can you understand every word without replaying it?
- Does the music vanish completely, or does it still support the scene?
- Are any syllables being masked by the beat or instruments?
- Does the voice sound clean at a low playback volume?
If the answer to any of those questions is no, adjust the mix before publishing.
A simple workflow is to test three listening modes:
- Headphones
- Laptop or desktop speakers
- Phone speaker
If the mix works on all three, you are in good shape.
Features and Benefits That Make Audio Workflow Easier
When you are creating lots of short-form content, audio balance is only one part of the job. The faster you can generate, revise, and publish, the easier it is to keep quality consistent.
That is where a tool like aicut can help.
Useful benefits for creators and marketers
- AI video generation, useful for turning scripts into ready-to-publish clips
- AI image stories, helpful for visual storytelling formats
- AI influencer videos, useful for UGC-style content and social proof concepts
- Motion control, for more polished visual pacing
- Viral prompt cloning, for recreating winning content structures quickly
- Campaign automation, for producing more variants with less manual work
- Multi-model access, for flexible content creation workflows
- Direct social publishing, for TikTok, YouTube, and Instagram distribution
The practical benefit is simple, you spend less time rebuilding the same type of content from scratch and more time testing what actually performs.
A Repeatable Order for Getting It Right Once
If you want a reliable process, use this order every time.
1. Choose the voice style first
Decide whether the narration should feel calm, high-energy, educational, or promotional.
2. Pick a track that leaves space
Avoid tracks with too much midrange clutter or distracting musical hooks.
3. Balance the voice at full clarity
The narration should already be easy to understand before any music is added.
4. Bring the music in low
Start lower than you think you need.
5. Adjust the music until it feels present, not dominant
The track should add mood, rhythm, and energy, without competing for attention.
6. Add ducking only if necessary
Use subtle automation rather than dramatic dips.
7. Test on a phone speaker
This final check catches the problems that studio monitoring often hides.
8. Export and listen once more
A final pass after export helps you catch any platform or codec quirks.
TL;DR, Key Takeaways
- How loud should background music be under a voiceover depends on the voice, the track, and the platform, not just one fixed number.
- A practical starting point is to keep music about 10 to 20 dB below the voice on average.
- Ducking usually works best when it is subtle, often around 3 to 8 dB.
- Set the voice first, then blend the music around it, instead of mixing the whole project louder.
- Always check the final mix on a phone speaker before publishing.
Conclusion
The best answer to how loud should background music be under a voiceover is not a single number, it is a process. Keep the voice clear, place the music underneath it with enough separation, and use ducking only when it improves clarity instead of creating obvious pumping. If you want to speed up your workflow while keeping your short-form videos polished, try aicut for faster AI-driven video creation, voiceover workflows, and direct publishing across TikTok, YouTube, and Instagram.