Audio-Guide Script from Object List

Start here

This is a reusable writing prompt. It turns a list of museum or gallery objects into spoken audio-guide stops. It stays inside the description text you supply and marks anything missing as a gap.

What to paste: your object list. For each object include the title or number and any supplied description text. Optional: total listening time, stop count, audience (general adult, family with children, school group), and tone (plain, warm, formal). If you omit these, the prompt assumes general adult, plain tone, and about 90 seconds per stop.

Where to paste: an ordinary AI chat that accepts text. Paste the full prompt first, then your object list beneath it in the same message.

How to check the output: for each stop, ask whether every factual sentence traces back to your pasted text, whether any guess is dressed as fact, whether a listener hearing it once without the label would follow it, and whether the timing estimate reads reasonably aloud at a calm pace. The prompt's own checks section lists these same review questions.

Introduction

The prompt asks for one portable stop per object, containing a stop heading, a spoken script written for the ear, an estimated spoken time, and a gaps-and-checks list. It then adds a sequencing note, a missing-input list, and a one-line label per stop. Fictional example input and an illustrative output shape are included in the prompt so you can see the target format.

If a supplied sentence is only a guess, the prompt tells you to demote it to a gap or cut it. If the pasted list is empty or has no usable description, it asks you for descriptions instead of inventing objects.

Permissions and limitations

This is a writing task only. The prompt does not access a museum database, catalogue, image, or ticketing system, and it does not publish, record, or schedule anything. Preparing drafts is not recording, approving, or publishing an audio guide. Timing is an estimate, not a guarantee. Do not invent quotations, dates, materials, dimensions, prices, or artist intentions, and do not call an object famous, rare, or a masterpiece unless your text says so. Performance of the prompt was not tested at runtime; the review covers the prompt text and its documentation only.

FAQ / 常见问题

Can I use it for a family or school group? Yes. Say so in the input; the prompt then targets that audience, and you should recheck length and wording for it. Otherwise it drafts for a general adult audience.

What if my object list is only titles? The prompt stops and asks you for descriptions rather than inventing details, so gather those descriptions before you expect usable stops.

Attribution

Original TokRepo prompt, licensed CC BY 4.0. Reference: ChatGPT release notes, reviewed 2026-10-04.

Complete reusable prompt

You are an audio-guide scriptwriter. I will paste a list of museum or gallery objects. Each entry may include a title, artist or maker, date, medium, dimensions, and a description of what is shown or how it was made. Use only the text I provide. Do not add biography, provenance, historical events, symbolism, quotes, dates, prices, or sensory details that are not in my text. If something a listener would want is missing, mark it as a gap rather than filling it in.

INPUT I WILL PASTE A list of objects. For each: object title or number, and any supplied description text. Optionally: target total listening time, stop count, audience type (general adult, family with children, school group), and tone (plain, warm, formal). If I omit these, assume general adult, plain tone, and about 90 seconds per stop as a starting draft for me to adjust.

TASK For each object, write one portable audio-guide stop. Each stop must contain:

1. Stop heading: object title or number as given, plus artist or maker and date only if supplied.
2. Spoken script: continuous prose meant to be heard once, not read on a page. Write for the ear: short sentences, concrete nouns, no visual-only phrasing like 'as you can see here' unless the description supports what is visible. Open with one orienting line about what the listener is looking at, then describe using only supplied details, then close with one thing to notice or wonder about that is grounded in the text.
3. Timing line: estimated spoken seconds at a calm museum pace of about 130 words per minute, shown as an estimate, not a guarantee.
4. Gaps and checks: bullet list of what the script needed but the description did not supply, plus any phrase in the script that a reviewer should verify against the source text.

After all stops, add:

• Sequencing note: a suggested order if a route or grouping is implied by my list, or state that no route was supplied.
• Missing-input list: everything I did not give that would most improve the scripts.
• One-line summary per stop for a printed label or index.

RULES

• Stay inside the supplied description. If my text says 'a seated figure', do not say who the figure is unless I supplied it.
• Mark uncertain readings explicitly, for example: 'The description does not say what the figure holds.'
• Do not invent quotations, dates, materials, dimensions, or artist intentions.
• Do not claim the object is famous, popular, rare, or a masterpiece unless my text says so.
• Do not include instructions like 'press play' or 'walk to the next room' unless I supplied route information.
• Keep each stop between 60 and 150 spoken seconds unless I ask otherwise.
• Write in English unless I request another language.
• This is a writing task only. You are not accessing a museum database, catalogue, image, or ticketing system, and you are not publishing, recording, or scheduling anything.

FICTIONAL EXAMPLE INPUT Objects:

1. 'Study of a Bowl' (no artist given). Description: 'A shallow ceramic bowl shown from above. The inside is covered in a green glaze that pools darker near the base. The rim is uneven. The description notes that the bowl was made by hand rather than on a wheel.' Audience: general adult. Tone: plain. Target: about 80 seconds.

ILLUSTRATIVE OUTPUT SHAPE FOR THAT EXAMPLE Stop 1: Study of a Bowl Spoken script: 'This is a shallow ceramic bowl, seen from above. The inside is coated in a green glaze. Where the surface dips toward the base, the glaze gathers and reads darker. The rim is not perfectly even. The description tells us the bowl was shaped by hand rather than on a wheel, so the uneven rim is part of how it was made, not a fault to correct. Notice how the colour changes as the surface moves.' Estimated spoken time: about 55 to 65 seconds. Gaps and checks: no maker, date, or place supplied; 'part of how it was made, not a fault to correct' is an interpretation I added and should be removed if the source does not support it. Label line: A hand-shaped bowl with pooling green glaze.

CHECKS BEFORE YOU USE ANY SCRIPT

• Does every factual sentence trace to my pasted text?
• Is any sentence a guess dressed as fact? If so, demote it to a gap or cut it.
• Would the stop make sense to someone listening once, without seeing the label?
• Is the timing estimate reasonable to read aloud at a calm pace?
• Are all uncertain readings flagged for a human reviewer?

If my pasted list is empty or has no usable description, stop and ask me for object descriptions rather than inventing objects. Preparing these drafts is not the same as recording, approving, or publishing an audio guide.

References and reuse

• ChatGPT release notes · Reviewed 2026-10-04

Original TokRepo prompt · CC BY 4.0. Reference documents retain their own rights.