How to write an audio guide script

An audio guide script template you can reuse

Use this four-part outline for one object. Replace every bracket with information checked against your collection records. Keep research notes and pronunciation instructions separate from the words visitors will hear.

• Locate: “You are beside [object and location]. Notice [one visible detail].”
• Explain: “That detail matters because [one verified fact or story].”
• Connect: “Compare [feature] with [another feature or a familiar experience].”
• Close: “Before you move on, look for [a final detail or question].”
• Production notes: record the object ID, source references, pronunciation notes, reviewer, approval date and measured read-aloud duration.

A short worked example

This invented example demonstrates the structure; it is not a description of an object at a Journey Pal client museum.

“You are beside a wooden travelling trunk. Look at the metal strips around its corners. They protect the places most likely to be knocked as the trunk is lifted and moved. Now compare the corners with the flat lid. Where can you see the most wear? Those marks invite us to imagine the journeys an object has made, but they cannot tell us who owned it. Before you move on, choose one detail you would ask its owner about.”

The opening locates the visitor, the next sentences direct attention, and the ending invites a response. For a real object, verify the construction and history before recording. Time the script aloud in the gallery, then revise it where the listener needs a pause.

Write for someone who is standing up

An audio guide is not an essay and it is not a label. Your listener is on their feet, holding a phone, looking at an object, and probably surrounded by other people. They can walk away at any moment and they will if the first sentence does not earn the next one.

That physical situation should shape every choice. Sentences run shorter than written interpretation. Clauses stack less. Names and dates arrive one at a time rather than in a cluster, because a listener cannot glance back at a sentence the way a reader can.

The most useful test is to read a draft out loud while standing and holding your phone. Anywhere you stumble, a visitor will stop listening.

How long each stop should be

Most successful stops land between sixty and ninety seconds, which is roughly 150 to 230 words at a natural speaking pace. That is long enough to tell one story properly and short enough that a visitor does not start wondering how much is left.

Length should follow the object rather than a house rule. A single striking artefact with one good story can justify two minutes. A room introduction whose job is to orient people should be closer to thirty seconds. Uniform length across every stop is usually a sign the script was written to fill a template.

If a stop runs past two minutes, it is almost always two stops. Splitting it also gives visitors a natural place to move, which keeps groups from bunching in front of one case.

• Room or gallery introduction: 20 to 40 seconds
• Standard object stop: 60 to 90 seconds
• Signature object worth lingering on: up to 2 minutes
• Children's trail stop: 30 to 45 seconds

Open with the object, not the context

The most common structural mistake is front-loading background. A stop that opens with two centuries of political history before mentioning the thing in the case has lost most of its listeners before the object appears.

Invert it. Name what the visitor is looking at, give them one concrete detail to find, then widen out into why it matters. The detail does real work: it gives the eye a job while the ear is busy, and it proves the guide is talking about this object rather than reciting a general history.

A reliable shape is: what you are looking at, one detail to notice, the story behind it, why it still matters. Four beats, one idea each.

Write it to be translated

If the guide will exist in more than one language, the English script is the source text for every other version, and choices that feel like personality in English can become expensive or meaningless elsewhere.

Idiom is the usual culprit. So are puns, cultural shorthand, sporting metaphors, and jokes that depend on English word order. A translator can handle them, but each one costs time and often produces a flatter line in the target language.

Plan for length changes too. Translated text commonly runs longer than English, and a stop that is already at two minutes in English can become uncomfortable once translated. Leaving headroom in the English version is easier than trimming eleven translations later.

• Avoid idiom, puns, and culture-specific shorthand
• Keep proper nouns consistent so every translator uses the same form
• Leave length headroom, because translations often run longer
• Mark words that should stay untranslated, such as place names and Māori terms

Say what a label cannot

If the script repeats the wall label, the guide has no reason to exist. Visitors can already read the label, usually faster than they can listen to it.

Audio earns its place when it carries things text handles badly. Voice can convey tone, doubt, humour, and pace. It can hold a pause. It can let a curator sound like a person with an opinion rather than an institution with a position. It can also carry sound that is not speech at all: a recording of the machine running, the room's acoustics, an oral history clip in the speaker's own voice.

A good question to ask of every stop is what this sounds like that it could not look like. If there is no answer, the stop is probably a label being read aloud.

Get it reviewed before it is recorded

Fixing a script is cheap. Fixing a recording is not, and fixing eleven translated recordings of a script that was wrong in English is the expensive version of the same mistake.

Build in a review pass before anything is voiced. Someone who knows the collection should check accuracy. Someone who does not should check clarity, because they will notice the assumed knowledge that specialists cannot see. If the guide covers cultural material, the relevant community should review it, and that review needs to happen while changes are still easy.

Read the final draft aloud, in the space, standing where the visitor will stand. Scripts that survive that test tend to survive contact with real visitors.