How to Add an AI Voiceover in Premiere Pro or DaVinci Resolve
Updated September 3, 2026
TL;DR: generate the voiceover in LeapFun (mastered WAV, SRT if you want captions), import it into your NLE, put it on its own audio track, align the first word to the first cut, split to land later lines, pull the music down or use a sidechain, export. Menu paths follow the current desktop versions and may differ by release.
Premiere Pro
- Import: File → Import (or drag the WAV into the Project panel). Import the SRT too if you exported subtitles; Premiere adds it as a caption track.
- Place: drag the clip to an empty audio track (A2 or A3), keep music on its own track.
- Align: snap the clip start to your first cut; use the razor (C) to split at sentence breaks and slide segments to later cuts.
- Levels: the LeapFun file is already at -14 LUFS. Leave it; drop the music 12–18 dB, or apply a ducking preset from the Essential Sound panel with the voice tagged as Dialogue.
- Export: Media Encoder with loudness normalisation off (the mix is already at target), captions burned in or as a sidecar.
DaVinci Resolve
- Import: Media page → drag the WAV into the Media Pool, or File → Import Media. Subtitles: File → Import → Subtitle to add an SRT track on the Edit page.
- Place: Edit page — drag the clip to a free audio track, rename the track "VO".
- Align: snap to the first cut; split (Ctrl/Cmd+B) at sentence breaks and move segments to hit later cuts.
- Levels: keep the voice track as is; on the Fairlight page lower the music bus 12–18 dB or add a compressor on the music with the VO track as sidechain.
- Deliver: export from the Deliver page; leave loudness normalisation off unless a platform demands a different target.
Tips
- Generate one file per paragraph for fast-cut videos so each line snaps to a cut.
- Export subtitles from Studio (SRT/VTT built from the narration timeline) instead of transcribing — they match the read word for word.
- If you need a re-take, regenerate one paragraph in LeapFun and replace one clip; don't re-import the whole file.
- For a music bed with automatic ducking and no DAW work at all, export the mix from Studio with your own track and skip the NLE audio pass.
Frequently asked questions
WAV or MP3 for the edit?
WAV. Both are mastered; WAV avoids a second lossy encode when the NLE exports.
Why is my voice quieter than the music?
The voice is at -14 LUFS, which is platform target; the music is probably peaking near 0. Pull the music down rather than pushing the voice up.
Can I get captions that match exactly?
Yes — export SRT/VTT from a Studio project; they're generated from the narration timeline, not transcribed.
Does this work with Final Cut or CapCut?
The steps are the same in any editor; see the CapCut guide for that app's menu paths.