Directing a Studio read: emotion tags, pauses and restraint

Most of what makes a narration good happens before you press generate. The choice of voice, a clean manuscript, sensible chapter breaks. The Studio tier adds one more tool on top of those: you can write short stage directions into the text, and the narrator will try to follow them. Used sparingly, they turn a competent read into a directed one. Used everywhere, they make a book sound like a radio play that has lost its nerve.

This article covers what Studio direction is, how to write it, and how to decide when a chapter needs any at all.

What Studio adds

VoiceCraft has three quality tiers. Quick and Standard share the four core narrators and are built for straightforward reading. Studio runs on a different speech model, unlocks a catalogue of thirty named voices you can hear on the voices page, and accepts inline emotion tags. On the pricing page, Studio costs the same per thousand words as Standard, so choosing it is a question of what the chapter needs, not of budget.

Every Studio chunk is also read with a standing instruction: a calm, measured narrator voice that pauses naturally at em-dashes. That default is worth knowing, because it tells you what you do not need to ask for. A Studio narrator is already calm. You only need to direct it when a passage should be something else.

How a tag works

A tag is a short instruction in square brackets, written into the text where it should take effect. The voices page gives two examples: [sigh] and [whispering]. A line like this:

"[whispering] Don't move. It hasn't seen us yet."

asks the narrator to drop into a whisper for that line. A tag like [sigh] placed before a sentence asks for the sound itself, then the words.

Think of tags as notes to a human reader in the margin of a script. They describe how the line should land, and a good reader interprets them rather than performing them literally. The model behaves similarly. The same tag can come out a little differently from one render to the next, which is one reason to listen before you move on.

Keep tags short and concrete. A single word or two that describes a delivery works better than a sentence of explanation.

Tags belong to Studio

Emotion tags are a Studio feature. Quick and Standard use a different speech model and do not list tag support, so a bracketed direction on one of those tiers may simply be read out as words. If you move a chunk from Studio back to Standard to save a render, remove its tags first. The same goes for your manuscript as a whole: if you plan to draft on Quick and finish on Studio, add the directions at the Studio stage rather than writing them into the first draft.

Studio voices are tier-locked too. A chunk rendered with one of the thirty Studio voices has to stay on Studio. Switch its tier and the voice clears, so you choose again. Our article on casting more than one voice explains how that works across a book.

Where direction earns its place

A small number of situations benefit most.

A shift the text cannot carry on its own. A character who whispers across three paragraphs of otherwise normal narration. A letter read aloud by someone close to tears. Where the page relies on italics or a line of description to set the mood, a tag gives the ear the same signal.

A beat the reader needs to feel. A single [sigh] before the last line of a chapter can do what a paragraph break does on paper. One such moment per chapter is usually plenty.

Dialogue with a clear emotional turn. If a conversation changes temperature halfway through, a tag at the turn tells the narrator where the change happens. Without it, the read can stay level straight through.

Where it gets in the way

Tagging every line of dialogue. A book narrated by one voice already carries its characters through phrasing and pacing. Directing each line makes the narrator sound busy and the prose secondary.

Asking for what the default already gives you. Studio reads calmly and pauses at em-dashes by default. Tags asking for "calm" or "slow" on ordinary narration rarely add anything.

Fighting the text. If a line is funny on the page, it will usually be funny read straight. A tag that tells the narrator the line is funny often makes it less so.

Tags in nonfiction. A practical guide, a history or a memoir that argues a case is best served by a clear, even read. Save direction for the rare quoted passage that genuinely needs it.

Pauses without tags

Some of the most useful direction does not need a tag at all. Because Studio pauses naturally at em-dashes, you can shape the rhythm of a sentence with punctuation. "She opened the door — and stopped." gives the narrator a breath before the turn. A full stop gives a firmer pause than a comma. A paragraph break gives more room still.

This works in plain text, costs nothing extra, and survives if you later move a chunk to another tier. Try punctuation first. Reach for a tag when punctuation cannot say what you mean.

A way of working that keeps costs sensible

Rendering is per chunk, and re-rendering a chunk costs credits again, so it pays to direct with a plan:

• Render the chapter once without any tags, and listen to it all the way through
• Mark the few moments that fell flat, usually a whisper that was read at full voice, or a turn in a conversation that passed without notice
• Add a tag or a piece of punctuation only at those moments, and re-render only the chunks you changed
• Listen again in sequence, not chunk by chunk, because a direction that sounds right on its own can sound odd next to its neighbours

Our article on working chapter by chapter goes further into how to spend credits on the chapters that carry the book. The pricing page shows the credit cost per thousand words for each tier, and the editor shows the cost of each chunk before you render it.

The short version

Studio direction is a fine brush, not a roller. Start from the default calm read, shape rhythm with punctuation, and add a tag only where the text needs a signal it cannot give by itself. Keep tags on Studio chunks only. Then listen in sequence. The chapters that sound most directed are usually the ones where you can barely tell anyone intervened.