Video Editing

How to Clean Up Voice Audio Before Adding Music

Clean the voice before you touch music: remove distractions, reduce steady noise, even out the level, then add music underneath. If you add music first, you'll keep turning it up to hide problems, and the voice gets harder to understand. Listen on headphones and on a phone speaker before you finish.

The order that works

  1. Cut distractions. Remove coughs, chair creaks, long breaths and mouth clicks between sentences.
  2. Reduce steady noise. Fans, AC hum and computer whine are constant, so noise-reduction tools can learn and reduce them. Use a light setting; heavy reduction makes voices sound watery.
  3. Even out the level. Bring quiet sentences up and loud ones down so the voice sits at a similar level throughout. Clip gain adjustments come first; a compressor can then smooth the remaining differences.
  4. Shape the tone (optional). A gentle low cut removes rumble below the voice. Don't chase perfection — clarity is the goal.
  5. Set a final level. A common target for online dialogue is peaks around −6 dBFS, with nothing hitting 0. Platforms normalise loudness differently, so consistency matters more than an exact number.
  6. Add music last. Bring it in well below the voice. Duck it (lower it automatically or by hand) whenever someone speaks.

A quick listening test

Play the edit on a phone speaker at medium volume. Can you understand every word without looking at captions? If not, the music is too loud or the voice needs more work.

Where to do it

  • Premiere Pro has the Essential Sound panel for dialogue cleanup and auto-ducking.
  • Audition gives more control over noise reduction and clip-level fixes — see the Audition guide.
  • DaVinci Resolve has the Fairlight page with noise reduction and dynamics.

Menu names and features change between versions; check your app's official help for the current names.

  • Distracting sounds cut
  • Steady noise lightly reduced
  • Quiet and loud sentences evened out
  • Dialogue peaks consistent, nothing at 0 dBFS
  • Music added last and ducked under speech
  • Checked on headphones and a phone speaker

Related lesson: Edit a clean 30-second talking-head sequence.

Sources and further reading

Try it in a lesson

Edit a clean 30-second talking-head sequence — A 30-second talking-head video with one clear point, no filler, tidy cuts and steady voice level.

Also in Video Editing: J-Cuts and L-Cuts for Beginners

Written and reviewed by the Step Education editorial team. Spotted something out of date? Let us know via the contact page.