Adding a voiceover (or voice over) to a video is two jobs. First you get the voice: record it yourself, or generate it from a script with an AI voice (text to speech). Then you fit it to the picture, and that part is the same whichever voice you chose. You split the script by scene, place each line where its scene starts, add captions that match, balance the voice against the music and the original sound, and listen through once before you export.
This guide walks through those steps in a way that works in any video editor, with the exact controls for iMovie on Mac and iPhone and for Clipchamp, as described in Apple’s and Microsoft’s own help pages. The last section shows how to have NoLang write and voice narration for footage you already have, and then fix it line by line.
| Step | What you do | Section |
|---|---|---|
| 1 Write the script | Give each scene of the video its own line or two | Step 1 |
| 2 Get the voice | Record yourself, or generate an AI voice from the script | Step 2a / 2b |
| 3 Line it up | Start each line where its scene starts and leave pauses | Step 3 |
| 4 Add captions | Match each caption to the line being spoken | Step 4 |
| 5 Balance the sound | Keep the voice on top of music and original audio | Step 5 |
| 6 Check | Listen through once and fix the worst spot first | Check before you export |
Your own voice or an AI voice
Every voiceover starts as one of two things: a recording of a person, or a voice generated from text. A professional voice actor is a third option, but the file they send back is placed on the timeline exactly like your own recording.
The choice comes down to what the voice has to carry and how often you will change the words.
| Record your own voice | Use an AI voice | |
|---|---|---|
| How it sounds | Your personality, energy and accent come through | You pick from a set of voices, and every video can use the same one |
| Fixing one line | Re-record it and match the tone of the lines around it | Edit the text and generate it again |
| Pauses and pace | Set by how you read; tightened or widened in the edit | Set by punctuation and speed settings; also adjustable in the edit |
| What you need | A quiet room, a microphone and a practice read | A finished script and one listen for misreadings |
| Best for | Vlogs, reviews, gameplay commentary, anything where who is speaking matters | Tutorials, explainers, training and videos you update often |
If you will revise the video often, use an AI voice. If it matters that the voice is yours, record it. You can also use both in one video. Put the switch on a scene change, so it comes across as a change of speaker.
Step 1: Write the script scene by scene
A voiceover drifts out of sync when its script is written away from the picture. Write the scene list first, then the lines, and only then record or generate anything.
- List the scenes: Play the video and note where each shot changes and what is visible in it.
- Give each scene one or two sentences: Describe what the viewer is looking at in that scene. A line about something that is not on screen makes viewers choose between watching and listening.
- Read it aloud and time it: Read each scene’s lines out loud at a normal pace and check that they fit inside the scene. If they don’t, decide now whether to cut words or hold the shot longer.
If the video teaches a task step by step, how to make instructional videos covers how to break the task into steps before you write a word.
Here is how the lines map to the picture in the example at the end of this guide, an 83-second how-to on repotting a houseplant:
| Scene | Time | On screen | Line |
|---|---|---|---|
| Hook | 0:00–0:05 | Roots hanging out of the bottom of a pot | Roots escaping from the bottom of the pot? Your plant is asking for a bigger home. |
| What you need | 0:17–0:22 | New pot, soil and tools on a table | Grab a pot one to two inches wider, with a drainage hole, plus fresh potting mix. |
| Step 1 | 0:22–0:28 | Watering the plant | Step one. Water the plant the day before. Moist roots slide out more easily. |
| Step 3 | 0:37–0:45 | Easing the root ball out of the pot | Step three. Tip the pot, support the base of the stems, and squeeze until it slides out. Never pull it by the stem. |
The full script is 219 words in 83 seconds, or about 2.6 words per second. That makes a useful first estimate: multiply a scene’s length in seconds by 2.5, and you have roughly how many words it can hold.
The same writing rules apply to a recorded voice and an AI voice:
- Keep sentences short, with one idea per sentence.
- Name things instead of pointing at them. “This” and “that” work on screen with an arrow, but not in a voice alone.
- For an AI voice, put punctuation where you want the pauses. In Clipchamp’s text to speech, for example, a period adds a moderate pause and a comma a short one.
Step 2a: Record your own voiceover
Three habits make a home recording cleaner and easier to fix, and none of them needs expensive gear:
- A quiet, soft room. Turn off fans and air conditioning, close the windows, and pick a room with curtains, carpet or a sofa to soak up echo.
- A steady distance from the microphone. Moving closer or further changes the loudness from line to line.
- Short takes. Record one scene or a few sentences at a time, so a slip costs one take instead of the whole read.
Set the input level before the first take. Apple’s iMovie guide puts it simply: keep the level meter green even when your voice is at its loudest. If the video’s own sound would play while you record, mute it so it does not leak into the microphone.
Many editors can record straight onto the timeline. Here is where that control sits in three common ones.
iMovie on Mac
- Place the playhead: In the timeline, move the playhead to where the voiceover should start, then click the Record Voiceover button below the viewer.
- Check the settings: Click the Voiceover Options button to choose the input source, set the volume of the microphone and, if you want, select Mute Project to silence the other clips while you record.
- Record: Click the Record button to start, and click it again (or press the Space bar) to stop. The recording appears as a new clip in the timeline, attached to the clip that was under the playhead. Click Done when you are finished.
iMovie on iPhone
- Place the playhead: Open the movie project and scroll the timeline so the playhead sits where the recording should begin.
- Open the recorder: Tap the Add Media button, tap Voiceover, and allow iMovie to use the microphone if asked.
- Record and keep the take: Tap Record; recording starts after a 3-second countdown. Tap Stop, then choose Review to listen, Retake to record again, or Accept to add it to the project.
Clipchamp
- Open the recorder: Open your video in Clipchamp and select Record & create, then Audio. Allow microphone access if prompted and choose the right microphone.
- Record: Select the red Microphone button, record, then select Stop.
- Keep or redo it: Preview the recording and select Retake recording or Save and edit. The recording is added to the timeline and to My media.
If you recorded on your phone or in another app, import the audio file into your editor and drag it onto the timeline under the video.
Step 2b: Generate an AI voiceover from text
Editors with text to speech usually work the same way: choose a language and a voice, set the speed and pitch, paste the script, preview it, and save it to the timeline. In Clipchamp it looks like this:
- Open text to speech: Select Record & create, then Text to speech.
- Pick the voice: Choose a language, then a voice. Select Hear this voice to preview it, and open Advanced settings to change the pitch or pace.
- Add the script and save: Type or paste the script, select Preview to listen, then select Save. The voice is added to the timeline and to My media.
Generate one clip per scene rather than one clip for the whole script. Each clip then starts at its own scene, and a fix touches only that clip. Clipchamp limits how many characters one text to speech item can hold, and Google Vids limits a script to 2,500 characters, so a long script has to be split anyway.
To change a line later in Clipchamp, select that clip on the timeline, open the text to speech tab, edit the text and save. If a name is read wrongly, spell it the way it sounds. If a number is read wrongly, write it out in full, so 1998 becomes “nineteen ninety-eight”. Then listen to the clip again.
You can also have the narration written from the footage itself. The NoLang section at the end of this guide shows how.
Step 3: Line up the voiceover with the picture
With the voice ready, put the video and the voice on the timeline and match them scene by scene. The buttons differ between editors; the work is the same.
- Keep the voice on its own track: Place the video on one track and the voiceover on an audio track below it. The camera’s own sound usually comes in attached to the video clip, so treat it separately from the voiceover.
- Start each line just after its scene starts: Use the start of each clip’s waveform (the shape that shows the sound’s loudness) as your guide, and place it just after the cut into its scene. A beat of picture first gives viewers a moment to take in the new shot before they listen.
- Split at sentence breaks to make pauses: Cut a long recording at the ends of sentences and move the pieces so there is a short gap between scenes. If you recorded or generated one clip per scene, the spacing between clips is the pause.
- Fix lines that run long: When a line is longer than its scene, hold the shot longer or shorten the line and record or generate it again. Speeding the voice up much sounds unnatural, so adjust the words or the picture first.
Which side gives way depends on the video. In talk-led videos such as explainers and commentary, the voice sets the pace and you cut the picture to fit it. In action-led videos such as cooking, crafts and repairs, the action sets the pace, so you fit the lines around the action and leave silence while something important happens on screen.
The example video is action-led. Most scenes move at a steady pace, but step four, the delicate one, runs at 95% speed with a 0.3-second pause after its opening line, and the last line is followed by 1.8 seconds with no narration over the final shot.
Step 4: Add captions that match the voice
Captions should say what the voice says, at the moment it says it.
- Start each caption when its line starts and end it when the line ends.
- Keep the caption text identical to the spoken line. If you change a line after captioning, change the caption too.
- Break long lines into shorter captions at natural pauses, so viewers are not reading ahead of the voice.
Automatic captions save most of the typing. In Clipchamp, turn on autocaptions from the captions tab, choose the language and select transcribe media; then correct any misspelled words in the transcript. The captions can also be downloaded as an .SRT file.
Step 5: Balance the voice, music and original sound
Once a voiceover is in, three kinds of sound overlap: the voice, the original sound recorded with the video, and background music. Keep them audible in that order.
- Decide what to keep of the original sound: Mute camera audio you don’t need, such as wind or room noise. Sound that adds to the scene, like sizzling or footsteps, can stay, turned down while the voice is speaking.
- Lower the music under the voice: Turn the music down until it no longer competes with any word of the voiceover. Between lines, it can come back up.
- Even out the voice: Lines recorded on different days or at different distances come out at different loudness. Set each clip’s volume so every line sits at the same level.
- Listen two ways: Play the video through laptop speakers and through earbuds. Music that sits nicely on one can bury the voice on the other.
Check before you export
Play the whole video once, start to finish, and check these six things.
| Check | What to look and listen for | What to adjust |
|---|---|---|
| Start and end | The first word isn’t clipped, and the video doesn’t end mid-sentence | Start and end of the voice clips |
| Scene match | Each line talks about what is on screen at that moment | Clip position, scene length |
| Pauses | Lines aren’t crammed together, and no line starts on the cut | Gaps between clips |
| Clarity | The voice isn’t buried under music or original sound | Music and original audio volume |
| Captions | Caption text and timing match the spoken line | Caption text and timing |
| Misreadings | Names and numbers are said the way you intended | Script spelling, re-record or regenerate |
Don’t try to fix everything in one pass. Fix the spot that is furthest off, then play it through again.
Add an AI voiceover to your own video with NoLang
NoLang can take footage you already have and add narration that fits what is on screen, with captions. Instead of placing audio clips on a timeline, you fix the result by editing the script and the voice.
Add narration to a video
- Open “From video”: On the Home screen, click See more, and under “How will you create it?” choose From video.
- Add the video: In “Add video and configure”, drop in a video or click “Select from past videos”. To replace the original sound with narration, check that the file shows the VIDEO ASSET label. In “Select from past videos”, a video chosen from the Upload tab gets this label. (The note on screen calls the same setting “Footage”.)
- Choose “Add narration”: In “Set generation method”, select Add narration. “Prompt to be sent” shows “Turn the attached video into a video with narration that fits its content. Do not use the original audio.” Click “Create using this method” and generate the video.
Fix the script, voice and volume after generating
- Rewrite a line: In the Reading Script, edit the text of a scene. That scene’s narration is generated again, and its captions change with it. For words that are read wrongly, use Pronunciation Correction to set how they are read.
- Change speed, loudness and voice: In the Reading Script, click the speaker icon on a scene and adjust Reading volume and Speech speed. “Apply to all scenes of this character” applies the change to every scene with the same voice. To switch to a different voice, choose Change character voice in the same menu.
- Lower the music: Right-click the background music on the timeline and use the Volume slider. Play the video afterward to check that the narration is easy to hear.
Start from your own recording
If you recorded the voiceover yourself, choose From audio instead of From video. Upload the recording and set the transcription language and the number of speakers; the video is built around your recorded voice, with captions of what you said.
Example: footage with an AI voiceover
This video puts an English AI voiceover and captions over how-to footage generated for the example. Each shot gets its own line or two, the on-screen list grows by one item as each is named, and the captions follow the voice. Use the template to build the same structure from your own footage and topic.
How-to footage with AI voiceover and captions
Explain how to repot a houseplant in seven steps, with an English voiceover and captions over the work footage.
If your team keeps producing how-to, product or training videos that need narration in the same voice every time, NoLang can also build narrated, captioned videos from a script, a PDF or a PowerPoint file. You can contact us to talk about using it across your company, or download the materials below.
Download Materials
