Audio Mark Duck

Automatically ducks the music bed under the original speech track using a sidechain compressor — the polished podcast or voiceover sound, without manual keyframing.

audio-mark-duck keys a sidechain compressor on the original audio track (the speech) and applies it to the overlay (the music), so the music dips when someone talks and recovers in the gaps.

Requirements. Ducking only works in audio-mark-mode=mix on a base that has audio. It is ignored in replace mode or on a silent base (there is no speech track to key on).

Each layer takes its own [N] suffix: audio-mark-duck[1]=0.8.