Your script, taken to the point of recording
Paste what someone is going to read aloud. Get the read prepared block by block with every acronym, number, date and two-pronunciation word decided; the voices cast with an engine choice and four settings each; the sound-effect cues written as generation prompts; and the music bed briefed with a ducking plan. The script reader runs in your browser and costs nothing.
Both examples ship with a saved run for every lane, so you can see all four outputs — including the artifacts — without signing in and without spending a credit.
Prepare the read
What this does, and what it does not
The script reader is a real parser, not a keyword search. Speaker labels are resolved per line
and only accepted when they are short and not a sentence, so So he said: does not
become a character while DR. REYES (clipped): does — and the parenthetical is kept
as a delivery note rather than thrown away. Bracketed stage directions are separated from the
spoken text, which is why the character count is the count of what a voice would actually be
sent. Sentences are split on terminal punctuation while protecting abbreviations, initials and
decimals, so Dr. Reyes paid $1.5m. is one sentence and not three. Everything the
reader finds is handed to the model as facts it must reconcile one by one, and the answer is
checked back against them — if the model skips a flag, this page says so.
It reads; it does not synthesise. No audio is generated here, no voice is auditioned, no music is rendered, and nothing is sent to any voice provider. This is the plan you then execute in whichever tool you use, and the prompt is told to describe voices by their characteristics rather than to name library IDs it cannot verify. A voice that imitates a real, identifiable person needs that person's recorded consent, and the prompt is told to say so rather than design around it. A token you paste is a token you should rotate: the reader flags it by line and the prompt is told to name it rather than repeat it.
Nothing to hand? Load the , a two-voice historical passage full of numbers, dates and units, or the , a thirty-second single-voice read with a URL, a price and a shouted line. Both replay a saved run in every lane for free.