In short: To add a voiceover in Premiere Pro, generate the narration in Speechma’s free Text to Speech and download it as WAV (use the small arrow next to Download). In Premiere Pro, choose File > Import or drag the file into the Project panel, drag it onto an audio track under your video and line it up. Then tag it Dialogue and the music Music in the Essential Sound panel, turn on Ducking, click Generate Keyframes and export.
Step 1: make the voiceover in Speechma
- Write for the earPlan about 150 words per minute of video. Write numbers the way they should sound (“twenty dollars”, not “$20”) and keep sentences short.
- Pick a voice and generatePaste up to 2,000 characters into Text to Speech, filter by language and country, preview a few voices and press Generate Audio. Adjust speed or pitch under Voice Effects if needed.
- Download as WAVPress the small arrow next to Download and choose WAV. WAV is uncompressed, so it holds up better when you edit, trim and add effects. MP3 works too if file size matters more.
- One file per sectionFor a long video, generate one clip per scene or paragraph and name the files 01, 02, 03. Fixing one line later means replacing one small file.
Step 2: add it in Premiere Pro
- Import the fileChoose File > Import and select your WAV, or drag the file from File Explorer or Finder straight into the Project panel.
- Place it on an audio trackDrag the clip from the Project panel onto an empty audio track (for example A2) below your video, so it does not overwrite the camera audio on A1.
- Line it up with your shotsMove the clip so the first word lands on the right shot. Cut the voiceover between sentences and slide each piece into place, or leave small gaps for breathing room.
- Tag the audio typesOpen the Essential Sound panel. Select the voiceover clips and click Dialogue, then click Auto-Match under Loudness. Select the music clip, click Music, and Auto-Match it too.
- Duck the musicWith the music selected, turn on Ducking and choose which audio types it reacts to (hover over the small icons to see them; pick dialogue). Adjust Sensitivity, Reduce By and Fades, then click Generate Keyframes. The music now dips under the voice and comes back up in the gaps.
- ExportUse Quick Export with H.264 and Match Source for a fast MP4, or open the full export settings if you need a specific size or bitrate.
Ducking settings: starting points
Every mix is different. These are safe starting values for a voiceover over background music; listen and adjust.
| Setting | What it does | Try this first |
|---|---|---|
| Sensitivity | How loud the voice must be before the music dips. Lower values react to quieter sounds. | Middle of the range, lower it if the music ducks too often |
| Reduce By | How far the music drops while the voice speaks | Enough that every word is clear on phone speakers |
| Fades | How quickly the music dips and returns, in milliseconds | Slower fades for calm videos, faster for punchy edits |
| Generate Keyframes | Writes the volume changes onto the music clip | Click again after any change (it replaces earlier keyframes) |
Setting names come from Adobe’s automatic audio ducking tutorial. If ducking is too strong, lower Sensitivity and click Generate Keyframes again.
Fixing common problems
- The voice and music fight each other. Make sure the voiceover is tagged Dialogue and the music is tagged Music, then regenerate the ducking keyframes.
- The voiceover is too long for the edit. Cut words from the script rather than speeding the voice up a lot. A small speed change in Speechma is fine; big changes sound rushed.
- A word is mispronounced. Spell it the way it sounds (for example “Speech-ma”), regenerate only that sentence, and swap the clip on the timeline.
- Your script is longer than 2,000 characters. Split it by paragraph and generate each part, or use SpeechmaPRO.com for 10,000 characters at once.
Example: a 60-second product video
Here is a short script that fits about 40 seconds of narration, leaving room for a logo and music at the end:
Meet the Trailhead bottle. It keeps water cold for twenty-four hours and coffee hot for twelve. The lid opens with one hand, so you never have to stop walking. It fits every car cup holder we tested. And when it gets dirty? Toss it in the dishwasher. Trailhead. Made for the long way home.
Generate it with a warm US English voice at default speed, download as WAV, and place it on A2. Put each sentence on its matching shot, tag it Dialogue, duck the music under it, and export with Quick Export.
Want the voice to sound like you?
With Voice Cloning you upload a clean 5 to 30 second clip of your own voice (or someone who gave you permission) and Speechma reads your script in that voice, in 10 languages. Need captions for the finished video? Make an SRT with our free subtitle generator.
Frequently asked questions
Should I use MP3 or WAV in Premiere Pro?
WAV is better for editing because it is uncompressed. In Speechma, use the small arrow next to Download and choose WAV. MP3 also imports fine if you need a smaller file.
How do I import audio into Premiere Pro?
Choose File > Import and pick the file, or drag it from File Explorer or Finder into the Project panel. Then drag it from the Project panel onto an audio track.
How do I lower the music under a voiceover in Premiere Pro?
In the Essential Sound panel, tag the voice as Dialogue and the music as Music. Select the music, turn on Ducking, adjust Sensitivity, Reduce By and Fades, and click Generate Keyframes.
Is the AI voiceover free to use in client work?
Yes. Speechma is free with no sign-up and commercial use of the audio is included, as long as you own the script.
Can I make the voiceover in other languages?
Yes. Speechma has 1,700+ voices in 75+ languages and accents, so you can make a Spanish or French version of the same edit.
How many words do I need for a 60-second video?
About 150 words per minute at normal speed, so roughly 130 to 150 words for a full minute, or fewer if you want music-only moments. Our voiceover length calculator can check it for you.
The Speechma newsletter
Make better voiceovers, faster
- New voices and features the day they launch
- Script templates and tricks that make AI voices sound human
- Creator stories, community picks and early access
