Voice Rehearsal
A full-duplex AI scene partner that reads your cues and listens for your lines.
Overview
Voice Rehearsal lets you run scenes out loud with an AI partner powered by the Gemini Multimodal Live API. The AI speaks every character's lines except yours, then listens in real time as you deliver your lines back. It's like having a scene partner available 24/7.
How It Works
- Select your scene and character on the setup screen
- Tap Go to connect to the AI session
- The AI reads the first cue line aloud
- Speak your line when it's your turn
- The AI evaluates your delivery, then continues with the next cue
- Repeat until the scene ends or you stop the session
The conversation is full-duplex — you can speak naturally without pressing any buttons. The AI listens continuously and responds in real time.
The On-Book Protocol
The AI follows strict rules to keep the rehearsal authentic:
| Rule | Behavior |
|---|---|
| Never volunteers your lines | The AI will pause and wait — it won't read your part for you |
| "Line?" prompt | If you're stuck, tap the Line button or say "line?" and the AI will read only your next line |
| Stays in character | The AI performs cue lines with natural pacing, not robotic text-to-speech |
| Corrects based on strictness | How strictly the AI corrects you depends on your strictness level |
Strictness in Voice Rehearsal
Your chosen strictness level controls how the AI responds to mistakes:
| Level | Name | Behavior |
|---|---|---|
| 1 | The Gist | Very forgiving — paraphrasing is fine; only corrects missed plot points |
| 2 | Rehearsal | Standard accuracy — minor word swaps are OK; corrects significant deviations |
| 3 | Opening Night | Word-perfect — interrupts on any deviation from the script |
You can change your strictness level in Settings or on the rehearsal setup screen.
Session Recap
When you end a session, your performance data is saved:
- Accuracy percentage for the session
- Trouble spots — lines you struggled with
- Session history visible in the Stats screen
Tips for Best Results
Quiet Environment
Background noise can interfere with speech recognition. Rehearse in a quiet space for the most accurate experience.
Use Headphones
Headphones prevent the AI's audio output from being picked up by your microphone, which improves recognition accuracy.
Mic Positioning
Keep your device within arm's reach with the microphone unobstructed. Avoid covering the mic with your hand or placing the device face-down.
Subscription Requirements
Voice Rehearsal requires one of the following:
- Voice Pro subscription (in-app purchase)
- Active production membership — If your stage manager has added you to a production in On Book Pro, Voice Rehearsal is included at no extra cost
Flashcard Mode is Always Free
If you don't have a voice subscription, you can still practice with Flashcard Mode at no cost.
Troubleshooting
| Problem | Solution |
|---|---|
| AI doesn't hear me | Check microphone permissions in your device settings |
| Audio is choppy or delayed | Ensure a stable internet connection — Voice Rehearsal requires real-time streaming |
| AI reads my lines instead of waiting | Make sure you've selected the correct character on the setup screen |
| Session won't start | Check your subscription status in Settings |
For more help, see the full Troubleshooting guide.
Related Guides
Last updated: February 19, 2026