Voice-Over Ducking Mixer

What ducking is and why every voice-over needs it

Ducking is the reason radio DJs, podcast intros and YouTube narrations sound effortless: whenever the voice speaks, the background music politely steps down; the moment the voice pauses, the music swells back. Without it you face an impossible compromise — music loud enough to hear drowns the words, music quiet enough for the words sounds dead in the pauses. Manual volume automation solves this but takes forever: a 20-minute narration can mean hundreds of tiny fader moves. Automatic ducking does the same job in one pass by letting the voice itself remote-control the music level. It's the same principle as sidechain compression on a mixing desk, just done with a clean digital gain curve instead of a compressor's detection circuit, which means no pumping artifacts when you set it up right.

Dialing in the three controls

Three numbers define the whole effect. Duck amount (default 8 dB) is how far the music drops while you speak — 6 to 9 dB suits most mixes; go past 12 dB and the music vanishes entirely, which can feel choppy under fast speech. Attack (default 60 ms) is how quickly the music ducks when your voice starts — fast enough that the first syllable is never buried, but not so fast (under ~20 ms) that you hear an audible click on each duck. Release (default 800 ms) is how gently the music returns in your pauses — this is the control that separates amateur from pro results. Too fast and the music "breathes" distractingly between words; 800–1500 ms gives that smooth broadcast glide. The voice gate (default -50 dB) simply decides what counts as "speaking" so room tone and mic hiss don't keep the music pinned down.

Getting the levels right before you duck

Ducking can't fix a bad balance — set your static levels first. Start with the music at full and the voice at full, then solo-listen: the voice should sit clearly on top with the music audible but ignorable underneath. A common mistake is ducking a music bed that's already too loud; if you need more than 12 dB of ducking, turn the music down instead. Watch the live duck meter while previewing: it should rest near 0 dB in pauses and dip smoothly to your duck amount during speech, never flickering rapidly. If it flickers, your gate is set too high and breath noise is triggering the duck — lower the gate threshold. Also mind the music choice itself: sparse, steady beds duck far more gracefully than busy tracks with vocals, which fight the narration no matter how you automate them.

From preview to finished file

The live preview uses the exact same ducking curve as the render, so there are no surprises — what you hear is what you get. When the music is shorter than the narration, the loop option repeats it seamlessly (pick a loopable bed for best results); when it's longer, the fade option glides it out over the final two seconds instead of cutting dead. Render length always follows the voice track, which is the part that matters. Export WAV when the mix is heading into a video editor or podcast DAW for further work — it's lossless and every tool on earth opens it. Export MP3 when this is the final deliverable for upload: 192 kbps stereo is transparent for voice-plus-music mixes and keeps file sizes small. Either way, keep your original voice and music files — re-rendering with tweaked settings takes seconds since everything stays on your device.