Skip to content

use case

verified July 2026

A narrator who never gets tired of take nine.

A video script is rewritten more often than it is recorded. On a meter every rewrite is a new bill, so the edit you should make becomes an edit you argue about. On a seat it is free, which is the only condition under which people actually iterate.

01

Why the meter shows up in the edit, not the invoice

The bill for a ten-minute video is small. The bill for a ten-minute video re-rendered eleven times as the script tightens is eleven times that, and it arrives in the same month you were trying to raise the quality. What the meter actually buys you is a reason to stop editing.

On a seat the eleventh render costs what the first one did, which is nothing at all. That changes what you are willing to try — an alternate cold open, a different pacing on the mid-roll, a second read of the same line at a lower intensity.

The edit changes at midnight

Film

The eleventh render costs what the first one did. What the meter really buys is a reason to stop editing.

02

What the API does here

  • POST /v1/tts/bytes returns one complete WAV — the whole voiceover track in a single request, ready to drop on a timeline.
  • Instant cloning means the narrator is you: five to ten seconds of reference audio, no enrolment step and no per-voice fee.
  • Inline controls place the pacing by hand — a break tag where the cut lands, a spell tag for a product code or a URL.
  • A seed makes a take reproducible, so a re-render after a script edit changes only the words you changed.

03

Notes — an engineer's checklist

01Can I clone my own voice for narration?

Yes. Five to ten seconds of your own recorded audio becomes a request parameter; there is no training job, no approval queue and no per-voice fee. The reference is fingerprinted and cached, so every render after the first skips the cloning work.

02How long can one script be?

Any length. A single request takes up to 2,000 characters, and long-form work is a loop over those requests — since characters are unmetered on a seat, a forty-minute script costs the same as a forty-second one.

03Will two renders of the same script sound identical?

Only if you ask. Send a seed and the same text, voice and parameters return the same audio, which is what lets you re-render a corrected paragraph and splice it into an existing track. Omit the seed for a fresh natural take.

See also — related sheets

A key and one seat to build this on — the same production API this page measures.

Request access for this use case