Runtime
Hear how long a scene really plays. At about 155 words per minute you can time acts, cold opens and monologues against your target.
For screenwriters and playwrights
An AI Table Read Before You Book the Room
An AI table read gives you a first listen of your screenplay, pilot or stage play without gathering actors. A narrator reads the scene headings and action, each character gets a voice, and you hear where a joke lands, where a speech runs long and which lines no person would actually say.
AI audiobook samples
View all
How to run an AI table read in four steps
Start with one scene to learn the format, then paste an act at a time. AI script reading works best in small, focused passes.
- 1
Copy the pages as text
Paste the scene as written, with a blank line between elements: a cue such as MAYA, an optional (parenthetical), then the speech. The table below shows each element.
- 2
Paste an act
Lines such as “Act Two” or a Markdown heading like “## Scene 4” start a new chapter, so you get one track per act or scene.
- 3
Cast the roles
Each character with a cue becomes a role. Audition voices, give the narrator a neutral read, and keep minor parts in the shared “Other voices” role.
- 4
Listen with the script open
Mark the lines that clunk, rewrite one in the studio and regenerate only that line to hear the fix before you change the draft.
Formatting a script for a virtual table read
Screenplay and stage play layouts are read as they are, so there is nothing to convert. Here is where each element of your script ends up.
| Script element | How to paste it | Who reads it |
|---|---|---|
| Scene heading | INT. KITCHEN. NIGHT. on its own line, as written | Narrator |
| Action | A plain paragraph, trimmed if you like | Narrator |
| Cue and dialogue | MAYA on its own line, the speech below it | Maya; the cue itself is not read |
| Parenthetical | (whispering) on its own line under the cue | The direction for Maya’s speech |
| Extension | THEO (V.O.) or MAYA (CONT’D) | The same role; the extension is dropped |
| Act or scene break | “Act Two” or “## Scene 4” on its own line | Starts a new chapter |
Stage play format works too, one speech per line: MAYA: We leave tonight. or MAYA (quietly): Not here. A character needs at least two lines to get a voice of their own, and any line can be moved to another role from its speaker menu.
What an AI table read lets you hear
Reading silently, your brain fills in the performance. Out loud, the page has to do the work.
Unspeakable lines
Sentences that read fine but tie a tongue in knots stand out at once, especially long setups before a punchline.
Voices that blur
If two characters sound interchangeable even with different voices, the difference is not on the page yet.
Exposition
Information dumps feel twice as long when spoken. You hear exactly where attention drops.
Button lines
Scene endings either land or trail off. A listen tells you which before an audience does.
Overlong action
Dense action paragraphs read aloud show where a reader of the script would start skimming.
Casting your table read from the voice library
Filter the library and audition voices for each role. A distinct voice per character makes it easier to judge whether the dialogue itself tells them apart.
81 voices
Who gets the most from AI table reads
A good fit
- Screenwriters revising a feature or pilot between drafts
- Playwrights preparing for a reading or a workshop
- Audio drama writers testing scenes before production; see the AI audio drama generator
- Writing groups who want to hear pages before they give notes
- Students in screenwriting and playwriting classes working on their own scripts
Not a substitute for
- Actors finding subtext, chemistry and timing in a room
- Casting decisions, since AI voices are not auditions
- A produced recording you plan to release
How an AI table read fits next to a live one
What it does well. It is ready whenever you are, at two in the morning or between drafts, and you can run it as often as you like. Hearing a scene five times while you rewrite it is a luxury no live table read can offer, because the actors go home.
What actors add. Performers bring interpretation, surprise and notes from the room. They find jokes you did not know were there. Use the AI pass to fix the problems you can hear yourself, so the live read is spent on the problems only people can find.
Directing the read. Parentheticals such as (dry) or (barely holding it together) become directions on their own, performed by the Standard, Lite and Expressive models. Add more in brackets, such as [over the noise], and inline tags like [laugh] or [sigh] add a beat inside a line. See AI voice emotion tags for the full list of cues.
Time and cost. A feature script of about 20,000 words, action included, runs a little over two hours aloud and uses roughly 1,000 credits on Standard or 700 on Lite. Check a scene in the words to minutes calculator first. Pasting and casting are free, and a Google sign-up comes with 100 gift credits (valid for 7 days, about 12 minutes on Standard; usage varies by text and voice tier) for a short sample.
Your pages stay private. Only the lines you generate in the AI audiobook generator are sent to the speech model, and scripts are never used to train AI models. Use your own scripts, or a writer’s script with their permission. Prose writers can do the same kind of pass with text to speech for writers.
AI table read FAQ
Can I upload my screenplay PDF for a table read?
Text PDFs open, but a PDF is reflowed into paragraphs, which can merge each cue into the lines around it. Export the script as plain text from your screenwriting app, or copy it from the page, and paste it instead. Cues, parentheticals and dialogue are then cast exactly as written.
Does the narrator have to read the action lines?
By default the narrator reads scene headings, action and stage directions, while cues are never read aloud. For a dialogue-focused pass, delete or trim the action and headings before you paste, and the cast plays the scene as pure dialogue.
How do I keep parentheticals like (beat) or (whispering)?
Leave them in. A parenthetical directly under the cue, or right after the name as in MAYA (whispering):, becomes the direction for that speech. Directions shape delivery on the Standard, Lite and Expressive models. The Premium model ignores them, so pick one of the others for a directed read.
How many characters can an AI script read include?
Up to ten named characters per project, ranked by how many lines they have, plus the narrator. Characters with a single line share the “Other voices” role. For a big ensemble, split the script by act so each project voices the people who matter in it.
Can I share the read with my writing group?
Yes. Download the AI table read as one MP3 per act or scene, or a ZIP of every track with a chapters index, and share it with whoever is giving notes. For a script you did not write, ask the writer before sharing any audio of it.
Hear your pages out loud tonight
Paste a scene and try an AI table read free for a short sample, no card needed.