Music too loud over the voice? How to balance them in a reel

Short answer: keep the music about 11 to 12 LU below the voice while someone is talking, bring it up when nobody speaks for more than 2 to 3 seconds, and start lowering it a little before the first word, not on it. Then check the mix on a phone speaker, not only on headphones.

I'm a sound engineer, and "the music covers the voice" is the note we get most often on short videos. Here's how we fix it in the studio.

1. Set the music by loudness, not by the fader

Two tracks at the same fader position can sound very different, because a dense pop track feels louder than a sparse piano piece. Measure instead of guessing:

• Measure the loudness of the voice alone (most editors show LUFS, or use a free meter plugin).
• Set the music so that, while the voice talks, it sits about 11 to 12 LU lower. If the voice is at -16 LUFS, the music sits around -27 to -28 LUFS under it.
• If there's no meter, a rough rule: lower the music until you can follow every word with your eyes closed, then lower it 2 dB more.

2. Let the music breathe in the gaps

A flat music level for the whole video sounds lifeless. Two rules:

• Short pauses (under about 2.5 seconds) between sentences: leave the music down. Pumping it up and down on every breath is tiring to listen to.
• Longer moments without voice (an intro, a b-roll section, the ending): bring the music up close to the voice level, about 2 LU below it. That's where the music gets to tell the story.

3. Get the timing of the fades right

This is what separates "edited" from "professional":

• Start bringing the music down about 0.7 seconds before the first word, with a ramp of about 1 second. If the dip lands exactly on the word, you hear it.
• Bring it back up with a quicker ramp, about half a second, once the voice is done.
• At the very end, let the music finish on its own ending or fade it over 2 to 3 seconds. A track cut mid-note sounds like a mistake.

4. Make room for the voice, not just a lower music

Turning the music down is half the job. The other half is EQ: the voice lives mostly between 250 and 500 Hz (body) and between 1.5 and 4 kHz (clarity). Dip those ranges in the music by 2 to 3 dB only while the voice is talking (a dynamic EQ or a sidechain does this). The music keeps its weight and its sparkle, and the voice sits in front without being louder.

5. The check most people skip

Listen to the final export on a phone speaker at medium volume. Phone speakers have almost no low end, so a bass-heavy track can disappear while a bright synth suddenly covers the voice. If you can understand every word on the phone and the music still feels present on headphones, you're done.

The common mistake

Using automatic ducking with a fast release. The music dives on every syllable and jumps back in every gap, and it sounds like it's breathing. If your editor has ducking, set the release to at least 300 ms, or draw the volume by hand.

If you'd rather not do it by hand

This is exactly the balance AudioReel applies automatically, together with voice cleaning and sound design, using the same numbers we use in the studio. But the rules above work in any editor.

FAQ

How loud should background music be under a voice-over?

About 11 to 12 LU below the voice while it speaks. Without a meter: low enough that every word is easy to follow, then 2 dB lower.

Should I use auto-ducking?

Yes, if you set a slow release (300 ms or more) and keep the music down through short pauses. Otherwise draw the volume by hand.

Why does my mix sound fine on headphones but not on the phone?

Phone speakers drop the low end and push the mids. Check on a phone before publishing and reduce the music's 1.5 to 4 kHz range if it covers the voice.