Screen Recording With Voice-Over: A Practical Guide

How to add a voice-over to a screen recording: write a script that matches the clicks, record it cleanly or generate an AI voice. Step by step.

A screen recording without narration is a puzzle: the viewer sees clicks and has to guess why. A voice-over turns the same clip into a lesson. This guide covers the two ways to get there (record your own voice or generate one), how to write a script that fits the footage, and the timing and caption details that separate a clean explainer from a rambling one.

Live narration or voice-over afterwards?

Talking while you record is fastest and sounds natural, but every mistake costs a retake and the pacing follows your mouse, not the viewer. Recording silently and adding the voice afterwards gives you a tight script, a quiet room and the option of an AI voice. For anything you will reuse (onboarding, docs, course lessons), voice-over afterwards wins. For a one-off reply to a colleague, just talk.

Step 1: Record the screen without worrying about audio

Follow the basics in how to record your screen: a clean window, a planned route, slow deliberate clicks. Leave a one-second pause after each important action so the narration has room. A silent recording is easy to redo in parts, so do not chase a perfect single take.

Step 2: Write the script to the footage

Watch the recording and write what the viewer needs to hear at each moment. Rules that hold up:

140 words per minute. That is a comfortable narration pace. A 90-second clip gets about 200 words. Count them.
Verb first. "Open Settings and choose Billing" beats "The next thing we are going to do is go into the settings area".
Say why once. One sentence of context per step, not a paragraph.
Name what is on screen. "The red Export button" lets the ear and the eye meet.
No filler. Delete "basically", "so", "as you can see".

Format the script with timestamps so you can match it to the video:

00:00  This is how to export last month's invoices.
00:04  Open Billing in the left menu.
00:09  Set the date range to last month and click Export.
00:16  The CSV lands in your downloads; open it in Sheets.

Step 3a: Record your own voice

Microphone: a USB mic or a wired headset. Laptop mics pick up the fan and the room.
Room: soft surfaces, closed door, no HVAC if you can help it. A closet of clothes is a studio.
Distance: a hand's width from the mic, slightly off-axis to avoid pops.
Levels: peaks around -12 dB. Too quiet is fixable; clipping is not.
Delivery: stand or sit upright, smile slightly, read in short phrases. Pause and restart a sentence instead of soldiering through a stumble; you will cut it later.

Record in one pass, then cut mistakes and tighten gaps. Normalize to about -16 LUFS for online video so it matches other content.

Step 3b: Generate an AI voice

Modern text-to-speech is good enough for product and training videos, and it removes the retake problem entirely: fix a word in the script, regenerate. What matters:

Pick a voice that fits the content. Calm and clear for documentation, warmer for onboarding. Listen to a full paragraph, not a five-word sample.
Punctuate for pacing. Commas and full stops become pauses. Use an ellipsis sparingly for a longer beat.
Spell out tricky words. Product names, acronyms and numbers: "A P I" and "two thousand twenty-six" often read better than "API" and "2026".
Keep sentences short. Long sentences flatten in TTS.

On takeascreen.com this is one panel: open the recording, click Generate script (it writes narration from the title, your prompt and the transcript, at about 140 words per minute for the video's length), edit, pick a voice and click Add voice-over. If the narration runs longer than the video, the last frame is held so nothing is cut. You can also ask the agent to make the whole video with narration from a single sentence.

Step 4: Sync the voice to the video

Two approaches:

Cut the video to the voice. Easiest when the voice is the backbone. Trim or hold frames so each on-screen action lands as it is mentioned. Holding the last frame is the standard trick when the voice ends after the footage.
Cut the voice to the video. Insert silences between sentences so the next line starts when the next click happens. Works well when the recording has natural pauses.

Either way, the rule is say it, then show it, or show it as you say it, never show it then explain it three seconds later.

Step 5: Mix system audio and voice

If the recording has its own sound (a video playing, UI clicks), decide whether it helps. Usually it does not: replace it with the voice. When it does (a product with audio feedback), keep it at roughly -12 dB under the narration so it reads as background. takeascreen.com's voice-over panel has a "keep original audio" switch that does exactly this mix.

Step 6: Captions

Captions are not optional in 2026: a large share of viewers watch muted, and captions make the video searchable. If you generated the voice from a script, you already have perfect captions. If you recorded, run a transcription and fix names. Burn captions into the video for social platforms; keep them as a separate track for players that support toggling.

Mistakes that make voice-overs sound amateur

Reading the UI aloud word for word ("Click the button labelled Save changes to save your changes").
Narration that ends ten seconds before the video does, leaving a silent tail.
Room echo. It is more noticeable than mic quality.
Different loudness between videos in a series.
Apologizing on the recording ("sorry, one sec").

Frequently asked questions

How do I add a voice-over to a screen recording for free?

Record the screen with a free online screen recorder, then record the narration with your phone's voice memo app in a quiet room and combine them in any free editor. Or use a tool that does both in one place.

Is an AI voice good enough for professional videos?

For documentation, onboarding, support and most product marketing, yes. For brand films where personality is the point, a human voice still wins. Many teams use AI for the long tail of how-to content and a human voice for hero videos.

How many words per minute for a voice-over?

About 140 for explanation, up to 160 for energetic marketing, down to 120 when the viewer must follow complex steps.

Can I change one sentence without re-recording everything?

With an AI voice, yes: edit the script and regenerate. With a human voice, re-record that sentence in the same setup and splice it in; match the mic distance and energy.

Keep going with how to record yourself and your screen if you want a face next to the voice.