Audio & mixing
Clean up dialogue and duck music
Use noise suppression, gate/expander, de-esser, and sidechain ducking carefully.
Start with the source#
Listen to an untreated section with speech, a pause, and a quiet ending. Cleanup controls are initially disabled so older projects remain neutral. Enable one process at a time and compare before adding another.
Noise suppression and gate#
Noise reduction uses lightweight six-band downward suppression for steady background noise. Set its floor conservatively to preserve quiet detail. It is not trained voice isolation or a repair tool for clipped recordings, clicks, wind, or missing audio.
A gate/expander can reduce the level in quiet gaps. An aggressive setting may cut off consonants, breaths, or reverberation. Check the start and end of each phrase instead of judging only loud speech.
Reduce sibilance#
The split-band de-esser reduces a targeted high-frequency region. Adjust while listening to words with strong “s” or “sh” sounds, then check that the voice has not become dull. EQ and compression may also change how much sibilance reaches it, so review effects order.
Make music follow dialogue#
- On the music channel, enable Music ducking.
- Select the dialogue track as the detector source.
- Set threshold and attenuation, then adjust attack and release for a natural response.
- Play sections with and without speech.
The detector listens before the source track's fader and processing; muting the source disables detection. Stem exports preserve the original detector input even when only the music is exported.