In this guide
Keep dialogue and music on separate tracks. Use music-clip volume keyframes for deliberate changes, or try Fairlight's ducker where available and choose the dialogue track as its source. Review quiet words, pauses and music recovery in the final mix.
To duck audio in DaVinci Resolve, keep dialogue and music on separate tracks and lower the music while the voice is speaking. For a few phrases, add volume keyframes to the music clip and shape the dip by hand. For a longer edit, try Fairlight's ducker on the music track, select the dialogue as its control source and adjust how far and how quickly the music recovers.
Ducking changes the balance over time; it does not remove music from a recording. Listen to the entire phrase and the spaces around it rather than choosing a setting from the waveform alone.
Choose a method
| Situation | Starting method | What to check |
|---|---|---|
| A few lines over a music clip | Manually shape the music clip's volume with keyframes. | First consonants, short pauses and the music's return. |
| Long dialogue with many turns | Try the Fairlight track FX ducker, where available. | Which dialogue source controls it and whether it responds to quiet speech. |
| Voice and music already combined | Look for original separate tracks; otherwise compare a separation attempt. | Missing words, retained music and processing artifacts. |
Prepare the dialogue and music
- Keep the originals. Duplicate the timeline or save a working copy. Put the voice and music on separate audio tracks and confirm that both are routed to the intended mix.
- Find a representative passage. Include a quiet phrase, a normal line, a short pause and a section without speech. Listen to the music on its own and then with the voice.
- Set a starting balance. Lower the music enough that the quiet phrase remains understandable. There is no universal dB value: the recording, arrangement, voice and destination all affect the right balance.
If you have multiple speakers, make sure the method responds to each one. A reference recording or another source may be cleaner than aggressive processing. Keep the original timeline timing if the audio is synchronized to picture.
Duck music manually with clip keyframes
Blackmagic's DaVinci Resolve 20 Beginner's Guide demonstrates adjusting the music clip's volume line and adding keyframes around speech. The method below follows that idea; controls may differ by version.
- Find the music volume line. In the Edit timeline, expand the music track enough to see the clip and its horizontal volume line. Play the passage and set a reasonable starting level.
- Mark the entrance. Place the playhead near the start of speech. Option-click on macOS or Alt-click on Windows on the music volume line to add points around the transition, as shown in Blackmagic's guide. Leave room for a gradual drop rather than cutting the music abruptly.
- Shape the dip and recovery. Add points around the end of the phrase and adjust the segment between them. The outer points preserve the surrounding level; the inner points define the quieter region. If the next voice begins quickly, keep the music low enough through the pause.
- Review in context. Listen from before the first word through the return. Move points or change the amount if the voice is masked, the music pumps noticeably or the transition feels abrupt.
For an ending fade rather than a dip in the middle, see how to fade audio in DaVinci Resolve. A fade handle at a clip edge and keyframes around dialogue solve different timing problems.
Use automatic ducking in Fairlight
Blackmagic's Fairlight overview describes a track FX ducker that uses one or more dialogue tracks to control a target such as music. The following is the documented approach; check the controls and feature availability in your installed version and edition.
- Open the Fairlight page and identify the music track and each dialogue track in the mixer. Keep the unducked mix available for comparison.
- Open the ducker on the music track and choose the dialogue track or tracks as its source in the ducker's selection menu. The source triggers the reduction; the music is the target being lowered.
- Adjust the duck level and recovery controls while listening. Start with a modest change, then check whether the music returns naturally between phrases without jumping up over the next speaker.
- Test every voice, a quiet line, a loud line and a pause. If a voice does not trigger the expected change, recheck the selected source and routing. Compare with the original mix and refine the controls or use manual keyframes for exceptions.
Fairlight also supports other dynamics workflows. Do not assume a setting shown for one version or project will give the same result in another. Use the current Blackmagic documentation for your installed release and save a copy before changing routing or automation.
What if music and speech are in one file?
A volume dip on the mixed clip lowers the speech and music together. First seek the original dialogue or music tracks. If only a finished mix is available, test a short representative section in the background-music removal workflow: describe the music and audition Background audio for the remainder. It is an estimate, not the original dialogue track; music can remain and speech can change.
The AI service uploads a supported MP3, WAV, MP4 or MOV up to 50 MiB. WAV downloads require an account and sufficient credits, with the cost shown in the result flow. Compare the quietest words and overlapping music before downloading. Keep the source start time and use the audio replacement checklist if you return the result to picture.
Troubleshoot and check the export
| Symptom | Check |
|---|---|
| Music does not duck | Confirm the ducker is on the music target and the intended dialogue source is selected; check routing and bypass. |
| Only one speaker triggers it | Review the selected sources and test each dialogue track independently. |
| Music surges between words | Adjust recovery or shape the short interval manually; listen through the full sentence. |
| Voice and music both become quieter | Check whether they are already mixed on the same clip or whether a control affects the wrong track. |
| Final file sounds different | Open the actual export, check included tracks and listen near the start, end and edited transitions. |
Check the result on headphones and a second playback system when possible. Listen at a comfortable level, confirm that quiet speech is intelligible, and review the whole export rather than a single transition.
Common questions
Does ducking remove background music?
No. It reduces the selected music level while the triggering voice or a manual automation dip is active. The music remains in the mix.
Can I duck music without a separate voice track?
You can draw volume changes by hand if you know where the dialogue occurs, but lowering a single mixed clip also lowers its voice. An automatic voice-triggered setup needs an appropriate separate control source.
What settings should I use?
There is no reliable universal duck amount or recovery time. Start gently and compare the same passage, including quiet speech and short gaps. Adjust until the voice is clear and the music returns naturally.