Write Commercial Voiceover for the Way It Will Be Spoken
Write Commercial Voiceover for the Way It Will Be Spoken
Read a commercial voiceover silently and you're checking it for the wrong things. Silent reading rewards a sentence that holds together: one subject, clauses in order, a tidy landing. Spoken copy is asked for something else. It has to arrive as a thought, in order, at a pace a person can be heard at, while the audience is also looking at something.
The answer is not to read faster, and it is not to pile on performance notes until the line sounds convinced of itself. Four moves, in order. Break the passage into units a listener can hold. Name the word each unit is standing on. Put the picture beside the words and find where the two are competing for the same half-second. Then read it aloud once, on purpose, as a scratch read: a check on the writing, not a rehearsal of the performance.
Who is speaking, and to whom, sits upstream of all of this. Once that's settled, what's left is the line's own problem — what it has to carry, and where it has to get out of the way.
Sentences hold together; thoughts land
A thought unit is the smallest stretch of speech that delivers one complete idea to the ear. Length is a symptom, not the definition. A short line can stack three thoughts with no room between them. A long one can stay a single thought the whole way and be perfectly speakable.
What decides it is what the listener has to hold in memory. On the page, "because" costs nothing — the eye slides back and checks. In the ear, "because" opens a bracket that stays open. If the second half arrives well after the first, the first has degraded by the time the bracket closes. That's what a crowded line feels like from the inside: pieces landing before the earlier pieces have settled.
Watch for the connectives that ask for memory: because, so, which means, not only, as well as, and then. None of them are wrong. Each one is a debt the listener has to carry, and a script full of them spends an attention budget that the picture is also drawing on. When the relationship matters — cause, contrast, sequence — keep it and shorten the distance between the two halves. When it doesn't, cut the connective and let two sentences stand.
Then say the unit out loud, at a pace someone could plausibly be heard at. If you have to break it to breathe, the break lands inside a thought, and the listener hears the idea split. Breath capacity varies between people, so this isn't a rule about syllables; it's a signal that the unit is doing more than one job.
Line breaks in a voiceover script are reading aids. They should mark where one thought ends, not where a rhythm would look nice on the page. A break invented for meter produces speech that sounds designed rather than meant.
Choose the word the line is standing on
Every spoken line has a word that takes the beat. If you don't choose it, the mouth chooses for you — usually the last content word, or the most concrete noun. That default is often fine. It's a problem when it isn't the word the spot is arguing for.
Two things drive the choice: what the line exists to say, and what the speaker knows at the moment of saying it. Someone who already knows the answer stresses the answer. Someone arriving at it stresses the arrival. Same words, different line.
Emphasis can also manufacture ambiguity rather than resolve it. Read evenly, "You made room for this" can drift toward "you made time for this" — made room slides toward the more common idiom when nothing presses on it. The picture can settle that, and so can the beat. But it has to be one or the other, and somebody has to decide.
Mark the beats lightly: underline, bold, or write the stressed word above the line. Those marks are reading aids for whoever reads next. Changing a word is a different act with a different owner, and in commercial work the difference matters more than it looks. Retyping a line is a proposal, not an edit.
The picture has a claim on the same attention
Build the scratch script in two columns. On the left, what's on screen. On the right, what's said. You don't need storyboard frames; the shot description is enough. What you're looking for is where a phrase sits over picture that's already busy, and whether the phrase is doing new work or narrating what the audience can see for itself.
A pause in spoken copy is not where the writer got tired. It's either breath or a decision to let the picture speak. If you can't say which, it will get filled in the edit, and probably by something worse.
The audience cannot inspect and parse at the same time. So the important picture and the important words want to live in different moments. When they land together, one of them loses, and it's usually the picture — because people are trained to follow a voice.
Two failures look like opposites and come from the same place. A rushed delivery hides a complicated thought, making it sound concise when it isn't. A slower reading exposes copy that has no clear relation to the image, because there's finally time to notice the mismatch. Both mean the words and the picture were never introduced to each other.
Two versions of one line
The spot below is invented, so that there's one fixed set of facts to argue about. It doesn't describe a real client, campaign, or product.
Thirty seconds. A shared workshop, end of day. One member clears a long bench: stacks the clamps, wipes the top with a rag, sweeps the offcuts into a bin. No dialogue. The bench comes back to bare wood. The camera holds on it. Someone arrives at the edge of frame, sets down a tool bag, unzips it. The spot has room for a couple of lines of narration and one line that isn't ours to reword: "You made room for this."
The dense version reads fine on the page. It's one sentence and a landing.
A membership here buys you a bench, a pegboard, a shelf, and something less obvious: the four o'clock clear-up, when someone — maybe you — wipes the sawdust off the top, stacks the clamps back on the pegboard and sweeps the corner, because a bench left ready is what makes tomorrow morning possible for whoever walks in next. You made room for this.
Sixty-odd words. It has a subject, a parallel list, a colon doing real work, and a short sentence at the end to close it. Nothing an editor would circle.
Now listen to what it's asking for. Because opens a bracket at word forty-one and doesn't close for fifteen words. The list of three items at the top asks the audience to hold "bench, pegboard, shelf" while a fourth thing arrives. The fixed line lands while the second person is walking in and unzipping a bag — the one beat the earlier narration has not already covered — so the words sit on top of a moment the audience is still reading.
The spare version spends its words differently.
A membership buys you a bench.
[The clear-up plays. No speech. The bench comes back to bare wood. The camera holds.]
[Someone arrives at the edge of frame, sets a bag down, unzips it.]
You made room for this.
Twelve words. The first line completes a thought and stops. The clear-up runs silent, which is what the picture is for. The arrival gets its own moment. Then the fixed line, unchanged in wording, now has something to point at.
Notice what happened to the middle. There's a tempting line here — "Somebody clears it at the end of the day" — that narrates what the shot is already showing. Cutting it is what buys the silence. The silence isn't leftover time; it's a decision about who speaks first, the voice or the picture.
Now the fixed line, in four readings. The words stay exactly as they are. Only the beat moves.
You made room for this. The line becomes an address to one specific person. After a shot of a single member sweeping alone, that reads as credit. If the preceding shot had shown someone else doing the work, the same beat would read as accusation. The stress doesn't carry the meaning by itself; it carries it in relation to what came before.
You made room for this. The deed, not the person. This suits a spot arguing that the space is the product of a decision rather than a thing that simply exists.
You made room for this. The cleared surface, literally. This is the reading the wipe-down supports, because the picture is making room in the plainest sense.
You made room for this. Points at the arrival. It needs the audience to already be looking there, which is exactly what the silent hold buys.
None of these is the correct reading in general. The choice depends on what the spot is arguing and what the speaker knows at that moment. A flat reading isn't automatically wrong either — some lines are captions, and captions shouldn't perform. But here, "this" is the one word in the script that points without naming: its referent is whatever is on screen at that instant, and the line only works if the picture got there first. A flat read spends it.
Four decisions stay open, and they belong in the margin rather than in the read:
- Whether "this" is spoken before or after the arrival lands. The wording doesn't change either way; the placement does.
- Whether "a membership buys you a bench" is accurate to the actual offer, which is a claim its owner has to sign off on.
- Whether the hold on the bare bench is long enough in the edit to survive — an editorial question, not a spoken one.
- Where the beat falls in the fixed line, which is a direction to be settled with whoever performs it.
The scratch read, and what it can't tell you
A scratch read is a temporary reading at a plausible performance level, in the proposed sequence, under roughly the conditions the track will eventually be heard in. Declared is the operative word. Say what it is before anyone hears it — scratch, working pace, judge the words — because without that label a client hears an unfinished performance and gives notes about the performance.
Take the two-column script and read the right column. Let your eye move to the left column where the picture is doing something. Four things are worth catching:
Where you run out of air inside a thought, and the idea splits. Where you stress a word you hadn't planned to stress, which usually means the sentence was built to land somewhere else. Where you stumble — the mouth finds pileups the eye sails past. And where a unit overruns the shot it's sitting under.
Recording it on a phone is fine, and often useful, as long as the recording stays evidence about the writing and doesn't quietly become a candidate track.
What it can't tell you matters just as much. It can't give you a duration. Word count is a budget, not a schedule: useful for planning the shape of a script, dangerous the moment you start believing it. There is no words-per-second figure you can apply from the page and trust. Planning figures get passed around, they vary by genre and performer and direction, and the number you measured is a range with your name on it, not the spot's timing.
It can't tell you what a performer will do. Your read is one mouth with your knowledge of the intent. The final length is a fact that arrives late, in the edit, with someone else's voice in it. Which is the real reframe: the copy isn't being written to fit one duration. It's being written so the thought survives a range of them. A line that only works at one speed is fragile, and the person who discovers that will be the one recording it.
A synthetic read has one narrow use. It will play the words back at a steady rate and let you hear whether the line is a mouthful. It won't tell you where the pause belongs, because it doesn't know what's on screen, or what the speaker knows, or which word the spot is arguing for. Never let a text-to-speech pass stand in for a performance direction.
What goes out of this pass
A membership buys you a bench.
[The clear-up plays. No speech. The bench comes back to bare wood. The camera holds.]
[Someone arrives at the edge of frame, sets a bag down, unzips it.]
You made room for this.
The emphasis is on this, and it only holds if the picture got there first. The silence the spot needs is the one between the bare bench and the arrival — the few seconds where nobody explains what we're looking at. Everything still unresolved stays in the margin, not in the read: whether the arrival lands before or after the line, whether the membership claim survives sign-off, and whether anyone has been granted the right to touch the fixed words at all.
Frequently asked questions
Why can a voiceover line read smoothly on the page but fail when spoken?
Silent reading rewards a sentence that holds together, but spoken copy must arrive as a thought, in order, at a pace a person can be heard at while the audience is also looking at something. Connectives like because, so, or which means ask the listener to hold earlier pieces in memory; when the second half arrives late, the first has degraded.
What is a thought unit, and is it simply a short sentence?
A thought unit is the smallest stretch of speech that delivers one complete idea to the ear. Length is a symptom, not the definition: a short line can stack several thoughts, while a long line can stay one thought. Say the unit aloud at a plausible pace; if you have to break it to breathe, the break lands inside a thought and the listener hears the idea split.
How should I choose the word a spoken line is standing on?
Two things drive it: what the line exists to say and what the speaker knows at the moment of saying it. If you do not choose, the mouth often defaults to the last content word or the most concrete noun. Mark beats lightly as reading aids. Changing a word is a different act with a different owner; retyping a line is a proposal, not an edit.
How does the picture affect pauses and pacing?
The audience cannot inspect and parse at the same time, so important picture and important words want to live in different moments. A pause is either breath or a decision to let the picture speak; if you cannot say which, it will get filled in the edit. A rushed delivery can hide a complicated thought, while a slower reading can expose copy that has no clear relation to the image.
Can word count or a text-to-speech pass tell me how long a voiceover will take?
No. Word count is a budget, not a schedule; no words-per-second figure can be applied from the page and trusted, and planning figures vary by genre, performer, and direction. A synthetic read has one narrow use: it plays words back at a steady rate and lets you hear whether a line is a mouthful. It cannot tell where a pause belongs because it does not know what is on screen, what the speaker knows, or which word the spot is arguing for.