Hearing dialogue out loud is the fastest way to find a line nobody would say. With AI voices, writers hit two snags: a character's voice changes from read to read, and a name or a word with two readings is mispronounced every time.

The problem

Rerolling a read until it says a name right costs credits with no promise it will. And without a fixed voice per character, the second read does not sound like the first.

How Scenaloom handles it

  1. On Media, open Table read. Each character gets a voice and a pace; lock them, and every read of the project keeps them.
  2. Add names and tricky words to the pronunciation list, up to 60, with how to say them. Your notes on the list are never sent.
  3. Before it runs, the read shows how many characters it has, about how many minutes of speech that is, how many parts and the price. It is priced by how long it takes to say, so a Chinese or Japanese script, which packs more speech into each character, can take more parts than an English one of the same length.
The table read panel for Last Ferry: Theo and Maya each with a voice, a pace and Locked, a narrator for the action, and Theo in the pronunciation list
Voices locked per character, with a pronunciation list.
A real table read of the demo scene. It opens with a spoken note that the voices are AI; Maya, Theo and the narrator each have their own voice.

The result

A read where Maya always sounds like Maya and names come out right the first time. The read begins with a spoken note that the voices are AI.

Questions

How is a table read priced?

Per part, by the estimated length of speech. On Voice Standard a part holds up to about 460 seconds; on Voice Pro up to about 600. The price is shown before you start.

Can a voice change between reads?

Not once it is locked. Unlock it to choose another.

Try it on a story. Start free with “Last Ferry”, the example on this page, or with your own work.

Start free with this example