Story to Script Rewriter: turn a post or article into a spoken short-form script

Paste a Reddit post, an article or any block of text and get back a script written to be read out loud: a hook that opens it, the narration itself, a title, two alternate hooks, the word count and roughly how many seconds it runs. You choose the tone, the platform, the point of view, the language and the target length.

It runs on your own OpenAI key, which you paste into the input. Without one the run still finishes, but it writes a clearly labelled canned sample instead of a real rewrite, so you can see the shape of the output before committing anything.

Input	A story, article or post. One, or a batch
Output	One script row per source
Ceiling	Each source is read up to 8,000 characters
Account needed	Your own OpenAI key for real output
Price	$0.02 per script, flat on every plan

🔍 What Story to Script Rewriter does

Reading well and holding a viewer are different jobs. A Reddit post can be a great read and still open with three sentences of setup that nobody watches past. This rewrites the same story so the turn lands in the first line and the rest earns its seconds.

Each source becomes one model call and one row. tone picks the voice: dramatic, casual, funny, emotional, educational or hype. platform shifts the pacing and the hook style for tiktok, youtube shorts, instagram reels or youtube. targetSeconds sets the length, worked as roughly 150 spoken words a minute. pov keeps the original voice or moves it to first or third person. language writes the output in whatever you name.

sources takes a batch: plain strings, or objects, in which case it reads selftext, then scriptText, then body, then text, then title. That last part matters because it means a Reddit scraper's rows go straight in with no reshaping.

📥 What you give it

{
  "text": "My neighbor kept parking in my spot. I asked nicely three times.",
  "tone": "dramatic",
  "platform": "tiktok",
  "targetSeconds": 45,
  "openaiApiKey": "sk-..."
}

Field	Default	What it is
text	none	One story. The Console prefills an example, but that is a prefill, so an API call has to send its own.
sources	none	A batch. Strings, or objects read in the order above. Added before text, so you can use both.
tone	dramatic	dramatic, casual, funny, emotional, educational or hype.
platform	tiktok	tiktok, youtube shorts, instagram reels or youtube.
targetSeconds	45	Aimed at, not enforced. It has no floor or ceiling, so a silly number gets a silly script.
pov	keep	keep, first or third.
language	English	Free text. The prompt wording is tuned for English, so other languages vary.
openaiApiKey	none	Your key. Secret field, never written to the log. Without it you get samples.
model	gpt-4o-mini	Any chat model your key can reach.
baseUrl	none	An OpenAI-compatible endpoint, if you are not using OpenAI's own.

Send something. With a key set and both text and sources empty, the run rewrites a built-in example story and that row is charged like any other. It is the one way to pay for a script you did not ask for.

Never paste your key into a published Apify task. Task inputs are readable by anyone who opens the page. Put it in the run input, or supply it per call.

📤 What you get back

Every run of this actor so far has been keyless, so what follows is the real labelled sample row from run T02XHkNtR8DzTbAbR, trimmed for length. With your key set, the same fields are filled by the model instead of the canned text, and _sample and _notice are not there.

{
  "ok": true,
  "source": "My neighbor kept parking in my spot. I asked nicely three times. ...",
  "title": "Sample Viral Short Script",
  "hook": "I thought this tiny mistake was over, but then the twist hit.",
  "script": "I thought this tiny mistake was over, but then the twist hit. My neighbor kept parking in my spot. I asked nicely three times. ...",
  "altHooks": [
    "This looked like a normal day until one detail changed everything.",
    "Nobody expected the ending to come from such a small mistake."
  ],
  "wordCount": 58,
  "estSeconds": 23,
  "_sample": true,
  "_notice": "SAMPLE OUTPUT - provide openaiApiKey for a real AI rewrite.",
  "tone": "dramatic",
  "platform": "tiktok",
  "requestedSeconds": 45
}

Field	What it is
script	The whole narration, hook included. Read it top to bottom and you have the voiceover.
hook	The opening line on its own, so you can test openings without splitting the script.
altHooks	Two more openings for the same story. Useful when you are posting the same piece twice.
wordCount, estSeconds	The actual length, worked at 150 words a minute. Compare it against what you asked for.
source	The first 2,000 characters of what you sent, so a batch row can be traced back.
tone, platform	Echoed back, which keeps a mixed batch sortable.

🧾 Reading the output

Rows land in two places, and which one tells you what happened.

Where	What it is	Charged
The dataset	A real script, ok: true	yes
Key-value store, SAMPLE_AND_NOTICES	A keyless sample, _sample: true with a _notice	no
Key-value store, SAMPLE_AND_NOTICES	A source that failed, ok: false with an error	no

So an empty dataset after a run with a key means every source failed. Open the key-value store tab and read the error on each one: it carries the status your provider returned, or a note that the model's answer could not be read as JSON.

One source failing does not stop the others, and the run still finishes green either way, so check the dataset row count rather than the run status.

▶️ How to run it

1. Open Story to Script Rewriter and click Try for free.
2. Run it once with no key and look at SAMPLE_AND_NOTICES to see the output shape.
3. Paste your key into OpenAI API key.
4. Put your story into Text / story, or a batch into Sources, then set the tone, platform and target length.
5. Click Start, then read the dataset or export it.

💰 How much does it cost?

$0.02 per script. Flat on every Apify plan, no volume tiers.

You pay per script row that reaches the dataset. Keyless samples and sources that failed go to the key-value store instead and are not charged as scripts. Your OpenAI usage is billed by OpenAI on your own key, separately from this.

💡 What people use it for

• Piping a Reddit scraper's rows straight into sources and getting a week of narration in one run.
• Testing three tones on the same story, then picking the one that reads best out loud.
• Turning a long blog post into a 45-second version for Shorts without rewriting it by hand.
• Building an A/B test out of hook and the two altHooks on a piece that underperformed.

🚧 What it does not do

• No voiceover, no video, no images. It writes text. The routing table below points at the rest.
• No fact checking. It rewrites what you gave it, including anything wrong in the source.
• No length guarantee. targetSeconds is an instruction to the model, not a limit, so check estSeconds on the row.
• Long sources are cut at 8,000 characters. Anything past that is not seen, so trim a long article to the part that matters.
• No retry. If your provider refuses one source, that source fails and the run moves on.
• No scraping. It does not fetch a URL. Send the text.
• The prompt is tuned for English. Other languages work, but the pacing advice is written for English rhythm.

🧭 Which short-form video tool do you need?

If you want	Use
A story rewritten into a spoken script	This one
A finished faceless short from a script	AI Faceless Short Generator
A score on how well a hook is likely to land	Hook Virality Scorer
Titles, descriptions and hashtags for the post	Social Metadata Generator
The Reddit posts to feed it in the first place	Reddit Scraper

❓ Questions people ask

Do I need my own OpenAI key? For real output, yes. Your key, your OpenAI bill, and nothing goes through ours.

What happens without a key? You get a labelled sample in the key-value store. Nothing reaches the dataset, so there is no script row to charge for.

Can I use a different provider? Yes, if it speaks the OpenAI chat format. Put its address in baseUrl and its model name in model.

Can I batch a whole subreddit? Yes. Put the scraper's rows into sources and every post becomes its own row.

Why is my script shorter than I asked for? targetSeconds steers the model, it does not bind it. Raise it, or send a richer source.

Where did my failed sources go? The SAMPLE_AND_NOTICES record in the run's key-value store, with the error on each one.

🆘 If something breaks

Open the Issues tab on the actor page. Send the run id and the tone, platform and target length you used. Never send your key. The error text in SAMPLE_AND_NOTICES usually names the problem by itself.