How to Write a Voiceover Script (With Templates)

A voiceover script is written to be heard, not read. Write one idea per sentence, keep sentences to about 20 words or fewer, and use contractions. Calculate the length from a words-per-minute rate: about 150 to 183 words make one minute of narration. Read the script aloud before you record it or generate audio.

Most scripts fail in the same ways. They run too long for the video, they use sentences that the speaker cannot finish in one breath, and they open slowly. The sections below cover length, writing rules, a simple format, four templates and the edit checklist.

How to write a voiceover script that sounds natural

Start with the outcome. Decide what the viewer should know or do after the video, then write the points in the order a listener needs them. Keep one main idea in each sentence. Use a plain word where a technical one adds nothing. Then read the draft aloud, mark every place you stumble, and rewrite those sentences first.

Your own point of view is what makes the script worth hearing. Use research, notes or experience that only you have, and treat any outside text as material to analyse, not to read aloud.

How many words does a voiceover need?

Length is set by the runtime and the pace. Two sources give the reference points. Brysbaert's 2019 meta-analysis in the Journal of Memory and Language found an average oral reading rate of 183 words per minute, across 77 studies and 5,965 participants. Rodero's 2015 study of 200 students heard roughly one-minute Spanish radio news clips. It judged 150 words per minute slow and 170 to 190 normal, with best recognition at 170 to 190. That study used Spanish radio, so treat it as a guide to English narration rather than a rule.

| Length | At 150 wpm | At 170 wpm | At 183 wpm | |---|---|---|---| | 15 seconds | 38 words | 43 words | 46 words | | 30 seconds | 75 words | 85 words | 92 words | | 60 seconds | 150 words | 170 words | 183 words | | 90 seconds | 225 words | 255 words | 275 words | | 3 minutes | 450 words | 510 words | 549 words | | 5 minutes | 750 words | 850 words | 915 words | | 10 minutes | 1,500 words | 1,700 words | 1,830 words |

These counts are calculated from the three rates. The 150 wpm column is a conservative planning figure, not a published average for English narration. It is widely repeated, but this guide did not find a primary study for it. YouTube's Shorts can run up to three minutes, so the Shorts maximum script is about 450 to 549 words at these rates.

Write for the ear, not the eye

Spoken copy has different rules from an article. The University of Florida's guide to writing for radio and television gives several rules that carry over to voiceovers:

Each sentence should be brief and contain only one idea.
Stick to sentences of 20 words or less, because the speaker has to breathe.
Write the way people talk. Fragments are acceptable if they make sense.
Use contractions.
Read the script aloud. The most important test is to hear it.

Other radio scriptwriting guidance adds that you should use commas and full stops, and avoid semicolons and colons, because they can disrupt the flow of a read. It also suggests avoiding long exact figures. "Nearly two thousand people" reads better than "1,988 people." The University of Florida guide also suggests spelling out numbers through eleven and using numerals from 12 to 999, and rounding unless the exact figure matters. These are broadcast-news conventions, so use them where they suit your format.

A simple voiceover script format

Most short scripts work in a single column, with a bracketed direction for the visuals and the spoken lines underneath. The example below is for a 30-second video about sourdough.

[Visual: a loaf on a wooden board, cut open]
Good bread needs three things. Flour, water and time.
[Visual: a jar of starter, bubbling]
Most of the work happens while you wait. You don't need a special oven.
[Visual: final loaf, close up]
Start with one jar, and bake your first loaf this weekend.

Keep the directions in brackets so they are easy to skip when you copy the spoken lines into a text to speech tool. Use a second column for longer videos if your editor works better with one.

Voiceover script templates

These editorial templates are structures, not facts. Copy one, then replace the content with your own material.

Explainer template (60 to 90 seconds)

A hook question in one sentence.
Why it matters, in one or two sentences.
Three points, each in two sentences.
A one-line recap.
A call to action.

Short-form hook template (15 to 30 seconds)

A bold claim or question in the first sentence.
One concrete detail.
The payoff.
A loop-back line that leads into the start again.

Product demo template (about 60 seconds)

The problem, in one sentence.
What the product or tool does.
One step-by-step moment, shown on screen.
The result.
The next step for the viewer.

Tutorial template (3 to 5 minutes)

An outcome promise.
What the viewer needs before they start.
Numbered steps, each with a one-line summary.
A common mistake to avoid.
A recap.

Writing for an AI voice

An AI voice reads what you give it, so the script carries most of the delivery. Write in short sentences with clear punctuation. Use full stops for pauses and commas for short breaks. Spell out numbers that the listener needs to hear clearly, and rewrite any name or term that comes out wrong when you listen back.

Describe the delivery in plain words, such as calm, warm or brisk, and add the supported expressive tags in the script. Listen to the full take, then revise the lines that sound wrong rather than adding more punctuation. Do not assume that one punctuation trick fixes every pronunciation. Test, listen and revise.

Cost is per character, so a tighter script also costs less. Onda Lite costs 1 credit per character and Onda Flash costs 1.5:

| Script length | Onda Lite (1 credit per character) | Onda Flash (1.5 credits per character) | |---|---|---| | 500 characters | 500 credits | 750 credits | | 1,000 characters | 1,000 credits | 1,500 credits | | 5,000 characters | 5,000 credits | 7,500 credits |

Hooks and the first 30 seconds

Hooks matter because viewers decide early. YouTube's "Intro" metric shows what percentage of the audience still watched after the first 30 seconds. The YouTube Analytics help page suggests that if the number is low, you should make the title and thumbnail match the content, and modify the first 30 seconds with different styles. The help page does not give a hook length, so do not treat any specific number of seconds as proven.

For a script, that means the opening should state the payoff clearly and get to the first useful point quickly.

Edit checklist

Each sentence carries one idea.
No sentence is too long to say in one breath.
Numbers are rounded where the exact figure does not matter.
The script is your own writing, not an article read aloud. YouTube's monetization policy does not allow readings of other materials you did not create.
The length matches your target runtime at your chosen pace.
You have read the script aloud once from start to finish.
The character count fits your plan and your credit budget.
Any dubbing plan leaves breathing room. Auto dubbing is unavailable when speech is too fast.

How OndaLabs fits

Paste the finished script into the text to speech tool, direct the delivery in plain words, add supported expressive tags, and listen. The Free plan's 2,000 credits, about 2 minutes, are enough to test a template. Starter, at $9 a month, gives 170,000 credits, about 189 minutes, Onda Flash and commercial use for published work.

For a two-speaker script, such as an interview or a host and guest format, the dialogue tool handles the two voices. Speech to text costs 50 credits per minute, so you can transcribe a recorded read-through and edit it into a cleaner script. OndaLabs has 37 languages with dedicated pages and 30 voices usable in every supported language. You translate the script yourself, and the voice reads what you provide. See pricing for the full plan list.

For the YouTube side of this process, read AI voiceover for YouTube. For short-form scripts, read TikTok AI voice.

Frequently asked questions

How do you write a voiceover script?

Start with the outcome you want, then write one idea per sentence in the order a listener needs it. Keep sentences to about 20 words or fewer, use contractions, and calculate the length from a words-per-minute rate. Read the whole script aloud before you record or generate the audio.

How many words is a 60-second voiceover?

At 150 words per minute, a 60-second voiceover is about 150 words. At 170 words per minute it is about 170 words, and at 183 words per minute it is about 183 words. Use the lower figure when you want a relaxed pace and the higher figure for a faster read.

What is the best voiceover script format?

A single-column format works well for most short videos. Put visual directions in brackets, and write the spoken lines beneath them, so the narration is easy to copy into a text to speech tool. Use a two-column layout for longer productions if your editor prefers it.

How do I write a script for an AI voice?

Write in short, clear sentences with plain punctuation, and describe the delivery in plain words. Add the supported expressive tags where they help, then listen to the full take and rewrite any line that sounds wrong. Keep the script in characters in mind, because cost is set per character.

How do I write numbers and abbreviations in a voiceover script?

Spell out numbers the listener needs to hear clearly, and round figures that do not need to be exact. Use "nearly two thousand" in place of "1,988" when the precise number is not important. Write out any abbreviation the listener would not recognise by sound, and check the result by listening to the take.

What is the average speaking speed in words per minute?

Brysbaert's meta-analysis found an average oral reading rate of 183 words per minute. A 2015 Spanish radio study judged 170 to 190 words per minute normal and 150 slow. Neither is a rule for every voice or language, so test your own script at the pace you want.