06 Auto-ducking

The room makes space for the voice. Then fills it back in.

Auto-ducking is what makes the difference between a podcast that sounds professional and one that sounds like a karaoke track. flexVox handles it with depth, attack, and release controls — plus an LUFS-aware mode that picks the right level on its own.

  • LUFS-aware
  • Three presets
  • Attack & release
  • Underlay + background
Dialogue · underlay · duck curve
Dialogue
Underlay · ducked
dialogue underlay duck curve depth −12 dB · attack 80 ms · release 280 ms

The thing that always sounded like the difference.

Listen to any professional podcast or radio drama. When the host talks, the bed drops a few dB. When they stop, it comes back up. That small move is what makes the dialogue feel present instead of drowning. It's also the move that almost nobody has time to do by hand, on every line, every episode. So flexVox does it.

Ducking is the difference between "I can hear them" and "I can hear what they're saying." The first one doesn't actually count.

Three controls. Three presets. One smart mode.

Auto-ducking exposes three parameters at the project level. The defaults are a sensible podcast curve. Three presets cover most other situations. And LUFS-aware mode hands the dialing over to the mathematician.

Depth

How far the underlay drops when dialogue is active. Default is around −12 dB; aggressive scenes go deeper.

Attack

How fast the underlay falls when dialogue starts. A short attack is responsive; a long attack feels relaxed.

Release

How fast the underlay comes back during dialogue pauses. Too fast pumps; too slow feels late.

The presets.

  • Gentle Shallow depth, slow attack and release. The bed stays present; dialogue sits on top of it. Good for warm conversational shows.
  • Default A standard podcast curve. Around −12 dB depth, ~80 ms attack, ~280 ms release. Where most projects start.
  • Aggressive Deep duck, fast attack. Bed almost disappears during dialogue. Right for high-stakes scenes, action, true-crime cold-opens.

LUFS-aware mode does the math.

Manual ducking lets you pick a depth in dB. LUFS-aware mode picks the depth for you, based on the actual loudness of both tracks. flexVox measures the integrated LUFS of dialogue, then measures the integrated LUFS of the underlay, and computes the depth that keeps them at your target separation — usually 9 to 12 LU.

The practical result: you can drop in a quiet ambient bed and a loud orchestral cue into the same project and both will sit at the right level relative to dialogue. No two-by-two dB-pushing required.

Edge cases the mixer handles for you.

  • Closely-spaced lines Dialogue ranges are merged before ducking so the bed doesn't pump in and out on every breath. Tight dialogue stays smooth.
  • Background music The episode-length music bed gets the same ducking treatment as underlay turns — one continuous curve across the whole show.
  • No underlay, no work Projects without underlay or background music don't run the ducking pass at all. It's free of cost when you're not using it.
  • Reproducible Ducking parameters are project-level. The same project always produces the same mix, with or without LUFS-aware mode on.

→ Keep going