A script written to be read is not a script written to be spoken. The same words, read aloud, become a different document: sentences that looked tidy on screen run out of breath, punctuation that looked correct disappears entirely, and names that were obvious in context turn into a guess.
Reformatting for voiceover is not cosmetic work you do at the end. It is the difference between a recording session that takes forty minutes and one that takes three hours and still sounds stiff.
Why formatting is not cosmetic
When someone reads a script aloud, they are doing three things at once: understanding the sentence, deciding how to deliver it, and watching for what comes next. Bad formatting makes all three harder, and the failures show up as the recording problems people blame on delivery — monotone, rushing, awkward pauses in the wrong places.
The fix is to move the decisions out of the recording session and into the document. Every mark of emphasis, every pause, every pronunciation should be decided while you are still alone with the text. Then the reader's only job is to say it.
Write for the ear, not the eye
Readers can re-read a sentence. Listeners cannot. So sentences that carry a subordinate clause inside another subordinate clause have to be split, even when the written version reads better as one sentence.
The practical rule is one idea per sentence. Not one clause — dramatic short sentences for their own sake are tiring — but one idea, so the listener never has to hold two unfinished thoughts at once. When a sentence includes a list, three items is the practical maximum before listeners lose the first one.
Two more adjustments matter. Replace constructions that only work on the page: parentheses become separate sentences and dashes become commas or full stops. And front-load the subject: listeners need to know what a sentence is about within its first few words, whereas readers can wait for the verb.
Punctuation that survives being read aloud
Standard punctuation is a guide for the eye and an unreliable one for the voice. A comma can mean a breath, a list continuation, or nothing at all depending on the reader, and a period is often read as a full stop when a short beat would sound better.
Two habits fix most of this. First, use the full stop aggressively. Short sentences give the reader somewhere to breathe and give the edit somewhere to cut. Second, spell out what you mean when the punctuation is ambiguous — write the pause rather than hoping the comma implies it.
Numbers and abbreviations need the same treatment. Write them the way they should be said: "twenty percent" rather than "20%", "about a thousand" rather than "1,000", and expand an abbreviation on first use unless it is genuinely read as a word.
Marking pauses deliberately
Pauses are the part of a voiceover that most scripts leave to chance, and they are the simplest thing to control. Adopt three marks and use them consistently.
- A short beat between related sentences — enough to separate them, not enough to sound hesitant.
- A hard stop at a section change, where the listener needs a moment to reset.
- A hold before a key phrase, to let the sentence that follows land with weight.
Mark them in the document, visibly, and not with symbols the reader has to interpret. Write "pause" or use a distinct bracket style rather than three dots, because a reader who has to decide what your notation means is no longer concentrating on the words. If the script is going into a recording session with someone else, agree on the marks in advance.
Timecodes and read speed
Read speed for narration lands in a fairly narrow range, and the useful way to use it is as a diagnostic rather than a target. Estimate by reading a paragraph aloud at delivery pace and counting the seconds per line; then use that to check whether a section fits its slot before you record it.
Timecodes in the margin — one per section, not per sentence — let you see the shape of the piece: where it is dense, where there is room, whether the ending arrives with enough time to breathe. Without them, sections run long by default and the shortening happens in the edit, where it costs the most.
If a section is consistently over its slot by more than a few seconds, cut content rather than speeding up. Rushing is the most audible defect in amateur narration, and listeners can hear a paragraph that is being delivered faster than it was written to be.
Difficult names and acronyms
Anything the reader might hesitate over should be marked in the script with a phonetic guide, even when it looks obvious to you. Names, product names, non-English words, and acronyms that are not pronounced letter by letter all qualify.
Handle acronyms by writing the pronunciation rather than the abbreviation after first use: if it is said as a word, write it as a word; if it is spelled out, space the letters so it is read that way. For numbers, note the intended reading when it is ambiguous — dates and version numbers are the usual offenders.
Keep a pronunciation list for your brand and product names and reuse it across scripts. Fifteen minutes once saves a re-record every time a new reader sits down, and it prevents a name being said three different ways across three videos.
The two-column script
For anything with visuals, split the page. Left column for what is heard, right column for what is seen, both aligned to the same timecode. This single change removes most of the ambiguity between the writer, the narrator and whoever is editing.
The rule that keeps it useful is that the columns should be independent: a reader should be able to read the left column straight through and get a coherent piece, and the editor should be able to work from the right column without needing to infer what the narrator will say. Where they conflict, revise the script rather than leaving both versions in the document.
A checklist before you record
Before anyone records, check six things. Sentences are one idea each and no list has more than three items. Every number and abbreviation is written the way it should be said. Pauses are marked visibly in three distinct grades. Section timecodes are in the margin and each section is within a few seconds of its slot. Every difficult name has a phonetic guide. And the accent or variety of the language is settled, because a script written in one variety and read in another produces constant small corrections.
None of this takes long, and all of it is cheaper than the alternative. A recording session is an expensive place to be making decisions about punctuation, and the takes where the narrator is working out how to say something are always the takes you do not use.