Audio ducking
Also known as: ducking, auto ducking, sidechain ducking
Automatically lowering the music volume whenever speech or another important sound plays, then raising it again when that sound stops.
Updated
What it means
Ducking lowers one sound to make room for another. In video, it usually means turning the music down while someone speaks, then bringing it back up in the pauses, so a voiceover or on-camera comment stays clear without the soundtrack disappearing.
It can be done by hand with volume keyframes, or automatically. Automatic ducking analyzes the speech track and lowers the music by a set amount whenever speech is detected; in audio engineering this is called sidechain compression. Most editors include an auto-duck feature for exactly this purpose, and many phone apps do it by default.
How it works in practice
Typical settings lower music by about 10 to 20 dB under speech, with fades of a few hundred milliseconds so the change is smooth. Shorter fades sound abrupt; very long ones let the music swallow the start of sentences.
In Premiere Pro, the Essential Sound panel tags clips as dialogue or music and creates duck keyframes automatically. Final Cut Pro, DaVinci Resolve and phone apps such as CapCut offer similar tools.
Ducking also works for ambient sound moments: dip the music when a wave crashes or a crowd cheers, so the real sound gets its moment. A two-second dip is usually enough before the track returns at full level.
What to watch out for
Pumping is the classic problem. If ducking reacts to every breath or background noise, the music bounces up and down constantly. Raise the detection threshold, lengthen the release, or duck manually on short sections.
Ducking cannot fix a speech track buried in wind noise. If the voice is not clear on its own, lowering the music will not help much.
Do not duck too deep. Music that drops to almost silent feels like a mistake. Keep it audible, just below the speech, and check the overall levels on headphones and on phone speakers.
How RawClip handles it
RawClip delivers a music-driven highlight with music included and cuts synced to the beat. If you add a voiceover afterwards in another editor, ducking the music under your voice keeps both clear.