What an AI scene partner actually does

“AI scene partner” can mean anything from a robotic text reader to a promise of a virtual co-star. This page describes exactly what the one in MasteringTheatre does, feature by feature, including its limits, so you can decide whether it belongs in your rehearsal routine.

Actors · acting teachers · 7 min read

The short version

An AI scene partner is a rehearsal tool that gives you the cue out loud, listens to your line, and tells you how close to the text you were. It replaces the friend who reads the other part when that friend is asleep, not the director, and not the actor opposite you on opening night.

What it does, step by step

1. It reads the script you give it

You import a scene as pasted text, a PDF or text file, or a photo of the printed page (photo import with text recognition is a Pro feature; on the free plan a single import is limited to 5,000 characters). Built-in sample scenes exist in every supported language if you just want to try. The app detects the characters, you select a passage in two taps and pick the role you play.

2. It speaks every role you do not play

Each other character gets a voice: the phone's built-in voices on the free plan, HD cloud voices on Pro, and on Premium a generative voice that receives a short written acting instruction from the server for the line. You choose a female or male voice per character and the reading speed (0.8× to 1.2× free, 0.5× to 2× on Pro). You do not choose the partner's emotion; there is no “play it angrier” slider. The partner reads. You act.

3. It waits, then listens

When it is your turn, on-device speech recognition transcribes what you say. Nothing leaves the phone for that step. In step-by-step mode (all plans) you tap “Validate” when you have finished the line. In fluid mode (Pro and Premium) a silence of your chosen length, from half a second to ten seconds, validates the line automatically, so a whole scene can run without touching the screen.

4. It scores the line

Each line gets a percentage: 90 % and above green, 70 to 89 % gold, below 70 % red. The comparison is phonetic and forgiving by design: homophones always count as correct, and on the default lenient setting close-sounding words pass. On Pro you can move the strictness to normal or strict, which refuses close-sounding substitutions. The score is about the words. It is not a rating of your performance.

5. It reviews the scene afterwards

During the scene, the app never corrects you out loud and never interrupts. At the end, the review lists every line with the words you missed or changed highlighted, and keeps the session in your history. Across sessions, an AI coach (Pro) comments on milestones: first time above 90 %, a weak line finally conquered, a personal best.

6. It can give acting notes, if you ask

Separately from scoring, the AI direction feature (Pro) analyses the scene and writes a note per line: intention, subtext, tone, one to three words to stress, a physical or rhythm suggestion. These are suggestions to argue with, not stage directions to obey. They are not stored beyond the session.

What it does not do

Where a real voice beats a synthetic one

If you always work with the same partner, ask them to record their lines once, line by line, into a voice pack. The app then uses their real voice as your cue when they are absent, and still listens and scores you. Voice packs stay on the device and can be shared as a file with the rest of the cast. Recording is free on every plan; rehearsing with a pack requires Pro.

What it costs

Rehearsing with system voices is unlimited on every plan; only AI features and cloud voices consume the daily energy credits.

This page is written from the app as it is today (September 2026). A score measures text accuracy, not talent. Published 2026-09-14.