The instrument
An AI scene partner, not a reader app.
Actor Insights is a line-memorization tool built for the way actors actually work. You paste your sides. You cast a voice for the other part. You run the scene. The partner speaks in-character using ElevenLabs v3; your voice is heard in real time via Deepgram Nova-3 streaming speech recognition. Your words vanish as you say them. When the cue word lands, the next line plays.
It is not a coach. It is not a notes-giver. It is a partner — one that lets you run the scene fifty times in a weekend without burning out the person sleeping in the next room.
What is Actor Insights?
A line-memorization instrument for working actors. Paste your sides or upload a PDF, cast an AI scene partner, and drill the scene word-for-word until it is under your feet. The partner speaks in-character, listens while you respond, and moves on when it hears your cue word.
How is this different from a regular line-learning app?
Most line-learning apps flash text at you. Actor Insights rehearses the scene with you. The AI actually speaks the other part — in a voice you cast for the character — and listens while you say yours. Your words vanish as you say them, so you cannot cheat by reading. That is the whole mechanic.
Is it good for audition prep?
Yes. Most actors use it for exactly this: you get your sides, you have three days, you need the scene in your body. Upload the PDF, cast a partner voice, run the scene five or ten times in a row on your phone, and the lines are somewhere permanent by the time you walk in the room.
Does the AI know when I am off-book?
It tracks per-line mastery across sessions. When you consistently land the cue words and the full line matches closely, that line is marked off-book. When you fumble it, the line resurfaces in the next drill. Curtain Call at the end of the session is the summary — what you landed, what is still wobbling.
Can I practice a public-domain scene without uploading my own?
Yes. The Practice Library has rehearsable two-hander scenes from Shakespeare (Hamlet, Macbeth, Romeo and Juliet, Midsummer), Chekhov (The Seagull, Uncle Vanya), Ibsen (A Doll's House), Shaw (Pygmalion), Wilde (The Importance of Being Earnest), and Sheridan (The Rivals). Each scene loads directly into the instrument.
What does the AI scene partner sound like?
Six pre-cast voices: June (steady, 30s, American), Bennett (classical, 50s, received English), Marcus (fast, 20s, Brooklyn), Stella (warm, 60s, American), Nico (light, 20s, Dublin), and Ruth (anchored, 40s, Nigerian). The parser infers which one fits your partner character automatically, or you can pick.
Does it work on a phone?
Yes — mobile is the primary canvas. 390×844 portrait. Press space on desktop, tap on mobile. Microphone access is required so the partner can hear you. There is a silent mode if you cannot talk out loud.
Is my script safe to upload?
Your scripts are private to your account. They are stored in Supabase with row-level security; only you can see them. We do not use them to train any model.