Mixing Music Beds: Ducking Levels Under Speech

A music bed sits under the voice and lowers itself whenever the speaker talks. That movement is called ducking, and it is the difference between music that supports a video and music that fights it. A fixed level that works in one scene will bury words in the next.

Ducking can be done with a sidechain, where the voice signal tells the music to lower, or by drawing volume automation by hand. Sidechain is faster, and automation is more precise. Most of my projects use a little of both.

How far to duck

A common starting point is to set the bed around 18 to 22 decibels below the dialogue when the voice is present. The exact number depends on the music, since a sparse piano tolerates less reduction than a dense drum loop.

Set the amount by listening on small speakers. If the bed still competes with the words there, it is too loud, no matter what the meter says. The voice has to win every time it opens its mouth.

The music bed is a supporting actor. The moment you notice it more than the voice, it has taken a role it was never cast for.

Set the timing carefully

Ducking should begin just before the voice, not after it, so the first syllable does not poke through at full music level. A short attack of maybe 50 to 100 milliseconds feels natural, and a slower release of a few hundred milliseconds avoids a pumping effect.

Release that is too fast makes the bed jump between phrases, which is more distracting than a steady level. Stretch the release until the music breathes back gradually and the motion stops drawing attention.

Do not lean on ducking alone

Ducking helps, but arrangement does more. A bed with fewer instruments in the vocal range leaves room without any processing at all. Choosing the right track beats fixing the wrong one.

I also ride the music through intros and outros, where there is no voice and the bed can stand forward. Letting the music lead in those moments makes the ducking under dialogue feel like a deliberate choice rather than a limiter.

Key takeaways

• Duck the bed 18 to 22 decibels under speech as a starting point, then refine by ear.
• Trigger the duck just before the voice so the first word never spikes.
• Use a slower release to stop the music from pumping between phrases.
• Pick beds with space in the vocal range so you rely on ducking less.