How to write an audio guide script that visitors listen to until the end

To write an audio guide script, write each stop for the ear: one idea, short sentences, an instruction to look, and a story instead of a list of facts. Keep it to about 90 seconds, read it aloud, and cut everything a visitor could read on the label beside the object.

This guide gives you ten rules, a structure that works for almost any stop, before-and-after examples and the mistakes that make people take their earphones out.

Writing for the ear is not writing a label

A label is read at the reader's pace. They can skim, reread a date and skip what bores them. Audio runs at the speaker's pace, once, in order. That changes everything about how you write.

• The listener cannot scroll back to check a name, so you repeat key words.
• Long sentences lose people halfway, so you break them.
• Visitors are standing in front of the thing, so you can point at it.

Visitors expect the audio to add something. In a study published in Curator: The Museum Journal (68:4), Kim and Kim (2025) report that "visitors strongly expected the audio guide to feature unique content rather than just a verbatim reading" of the label. If your stop repeats the label, you have spent 90 seconds of their visit on something they could read in ten.

The ten rules

1. One idea per stop. Decide the single thing the visitor should remember. Everything else is optional.
2. Start with a hook. The first sentence decides whether they keep listening. Open with a question, a surprising detail or a person.
3. Tell them where to look. "Look at the hand on the left." "Find the crack running down the wall." Without it, people stare at their phone.
4. Use short sentences. If you cannot say it in one breath, split it.
5. Prefer verbs to nouns. "The monks built the cellar in a single winter" sounds better than "The construction of the cellar was completed by the monks."
6. Use dates sparingly. One date per stop is plenty. "Two hundred years before the cathedral" is easier to hold than "1043".
7. Speak to the listener. "You are standing where the kitchen fire was" puts them in the scene.
8. Repeat the key word. Saying "the altarpiece" twice is fine. In audio, repetition is clarity.
9. End with a bridge. Send them on: "Now turn around and find the same saint in the window."
10. Keep it to about 90 seconds. Around 180 to 220 words. Longer stops belong in an optional second layer.

The five-block structure

Most good stops follow the same shape. Use it as a starting point, not a template to fill.

The audio guide script template turns this into a sheet you can fill in for every stop.

From catalogue entry to story

Here is a typical catalogue entry for an imaginary portrait, and the same content rewritten for the ear.

Before (catalogue):

Portrait of a merchant, oil on panel, c. 1530. The sitter, identified by the coat of arms in the upper right corner as a member of a leading merchant family, holds a letter, a common attribute denoting commercial activity. Recent technical analysis has revealed an earlier composition beneath the visible paint layer.

After (audio):

Look at his hands. One holds a letter, folded, with the seal still on it. In 1530, a letter like this could carry news of a ship lost at sea, or a fortune made. This man was a merchant, and his family was one of the richest in the city. You can see their coat of arms in the top right corner. But here is the strange part. When the museum looked at this painting with infrared light, they found another face underneath. The painter had started a different portrait, and covered it. Who was the first man? Nobody knows. As you walk on, look for the same family crest in the next room.

The facts are the same. The difference is the order, the instruction to look and the question that keeps people listening.

Who is speaking?

Decide on one voice and keep it.

• A narrator is the safe default: warm, clear, never showing off.
• A curator or an expert adds authority and personality, and works well for a few stops in their own voice.
• A character (a monk, a winemaker, a child of the house) can work for families, but needs care so it does not sound like a school play.

Mixing voices is fine if it is deliberate: a narrator for most stops and the head gardener for the garden stops, for example.

Directing the eye

The most powerful thing audio can do is make people look more closely. Build at least one instruction to look into every stop, and place it early.

• Name a position: "on the left", "near the floor", "above the door".
• Name a detail: "the dog asleep under the table".
• Ask them to move: "take two steps back and look at the whole wall".

Then give them a second to do it before you go on. In the script, a short sentence after the instruction does that naturally.

Test on the floor

A script is not finished until it has been heard in the room.

1. Read it aloud at the speed you will record it. Time it.
2. Stand where visitors stand and check that every "look at" points to something they can actually see.
3. Ask a colleague who does not know the object to listen and tell you what they remember. If it is not your one idea, rewrite.

Common mistakes

• Reading the label aloud. Visitors can read; give them what the label cannot say.
• Too many facts. Five dates and three names in 90 seconds leave nothing behind.
• A slow start. "This painting was created in…" loses people in the first five seconds.
• No instruction to look. The visitor ends up watching their screen instead of the object.
• Jargon. "Sfumato", "corbel" and "terroir" need a plain explanation the first time.
• Writing for specialists. Experts will forgive simplicity; beginners will not forgive jargon.

Writing for different audiences

The same object often needs two versions: one for adults and one for families, or a standard stop and an audio description for blind and low-vision visitors. Write the standard version first, then adapt. Writing audio guides for children and families explains what changes.

Writing in paragraphs

Write each stop as a few short paragraphs rather than one block. Paragraphs make the text easier to edit, to translate and to record again one piece at a time. In Muvo, each stop is a list of paragraphs of up to 800 characters, and each one can be recorded, voiced or translated on its own.