How to prep a voice-over script that reads clean

A well-prepped script is the difference between a 20-minute session and an hour of retakes. Here is exactly how we format, mark and time a script before anyone steps into the booth.

Most voice-over sessions that run long don't run long because the talent is slow. They run long because the script arrived as a wall of marketing copy with no pronunciation guidance, no tone direction, and a runtime that turns out to be twice the slot it has to fit. Every one of those problems is fixable at the desk, before the red light goes on. Here is the exact prep we do on every script that comes through the studio.

1. Read it out loud before anyone else does

This is the single highest-value step and the one most people skip. Copy that scans perfectly on a slide can be genuinely hard to say. Stacked adjectives ("a powerful, intuitive, all-in-one platform"), long subordinate clauses, and clever parentheticals all read fine silently and turn into tongue-twisters aloud.

Sit down and perform the whole thing at speaking pace. Every place you stumble, hesitate, or run out of breath is a place the talent will too — except they'll do it on the clock. Mark those spots. Usually the fix is simple: break one 34-word sentence into two, or swap a written-register word ("utilise", "commence") for the way people actually talk ("use", "start").

If you can't say it smoothly on the first try in your own voice, don't expect a stranger to nail it in the booth.

2. Build a pronunciation key

Every brand name, place name, acronym, product name, drug name, and foreign word needs a note the first time it appears. Don't assume it's obvious — "Yuntaku" alone has at least three plausible readings. Spell it phonetically in brackets right after the word, using plain syllables and capitals for the stress:

• Brand: Yuntaku [yoon-TAH-koo]
• Acronym spoken as letters: API [say each letter: A-P-I]
• Acronym spoken as a word: NASA [say as a word: NASS-uh]
• Numbers: 2024 [say "twenty twenty-four", not "two thousand and twenty-four"]
• Ambiguous: "read" [past tense: RED]

This is the single most common cause of pickups — the extra session where you re-record a handful of lines because a name was said wrong. A five-minute pronunciation key at the desk saves a whole scheduling round later.

3. Direct the tone, section by section

One line of direction at the top of a section does more than a paragraph of adjectives buried in an email. Write it as if you're briefing an actor, because you are: "Warm, unhurried, like explaining something to a smart friend over coffee." That tells a talent exactly where to sit emotionally.

Where the energy needs to shift — say a documentary that moves from a somber history section into an upbeat present-day section — mark the shift explicitly at the transition. Don't make the reader guess where the mood changes; the guess will cost you a take.

4. Format the page so it can be performed

A script is a performance document, not a printout. Format it to be read while standing, at a mic, glancing down between lines:

• Generous line spacing (1.5 minimum). A dense block is hard to track when your eyes keep leaving the page.
• One idea per paragraph so natural breaths land at paragraph breaks instead of mid-thought.
• Bold the words that carry the emphasis. "This is the only tool you'll need" reads very differently from "This is the only tool you'll need." Don't leave it to chance.
• Mark deliberate pauses with a slash (/) for a short beat, double slash (//) for a longer one.
• Number your sections or paragraphs if there's any chance of recording out of order or doing pickups — "let's redo paragraph 7" beats "the bit near the middle about pricing."

5. Time it to the slot — before you record

The most expensive surprise is discovering in the edit that a 30-second script is actually 41 seconds of copy. A comfortable, natural read for most narration is roughly 150–160 words per minute; energetic commercial reads push to 170–180; measured e-learning sits around 130–140. So a 30-second spot wants somewhere near 75–80 words, not 110.

Count the words, divide by your target rate, and check the math against the slot before recording. If it's long, cut at the desk where you have time to think, not in the booth where you're tempted to just speed up the read and lose all the warmth. Rushing copy to fit is the number-one reason a spot sounds "salesy" instead of confident.

6. Attach the boring but critical metadata

The script should not travel alone. Along with it, send:

• Target runtime and the exact slot it fills.
• Delivery format — WAV vs MP3, sample rate, mono/stereo, and any loudness target (e.g. −16 LUFS for web, −24 LUFS for broadcast).
• Intended usage — broadcast, paid social, internal training — because it affects both the read and the licensing.
• Any legal or disclaimer lines that must be included verbatim, flagged clearly so they don't get trimmed.

Nothing derails a delivery like discovering after the session that a mandatory 8-second legal tag was never in the script.

The pre-session checklist

Before a script is "booth-ready," it should pass all of these:

1. Read aloud end-to-end, stumble points fixed.
2. Every proper noun and acronym has a phonetic note.
3. Each section has a one-line tone direction.
4. Emphasis and pauses are marked.
5. Word count checked against the runtime at a realistic WPM.
6. Delivery format, usage and legal lines attached.

None of this is glamorous, and that's the point. A clean script turns the recording into a smooth run instead of a negotiation — and it's the cheapest quality upgrade available to any project, because it costs nothing but attention.

Got a project? Request a demo and we'll send sample reads.