Getting Started

1. Start the server — run ./start.sh in terminal.
2. Paste your script into the text area.
3. Click "Enhance with AI" (optional, needs API key) — adds color-coded reading cues.
4. Click "START PROMPTER" — opens the teleprompter and starts voice tracking.
5. Click "End & Review" when done — get a delivery score with detailed feedback.

Visual Cues (from AI Enhancement)

Each cue tells you how to deliver that word. Cues appear on upcoming words and fade once spoken.

Live Coaching

While you read, the app analyzes your microphone in real time:

• Energy Meter (toolbar) — shows your current volume level.
• Coaching Alerts (center overlay):
  • ⚠️ Monotone — pitch isn't varying enough
  • 🐢 Too fast — slow down
  • 📢 Low energy — speak up
  • 🌊 Flat volume — vary between softer and louder
  • ⏸ No pauses — take a breath between points
  • 👏 Great energy — positive reinforcement
• Performance Dashboard (press S) — live star ratings.

Post-Session Review

Click "End & Review" when done. You get:

• Overall Score (1-10) with color coding
• Category Scores — pace, energy, clarity, pauses, filler words, accuracy
• Highlights — what you did well
• Improvements — specific, actionable tips
• Key Moments — timestamped good and improvable moments
• Transcript — exactly what Deepgram transcribed

Keyboard Shortcuts (Prompter Mode)

• Space — pause / resume
• ↑ ↓ — manual scroll
• ← → — jump 5 words
• + - — font size
• L — legend
• S — stats
• H — help
• F — fullscreen
• ? — keyboard shortcut cheatsheet
• ] — skip word   \ — skip sentence
• Esc — back to editor

Tips

• Use a decent microphone in a quiet room.
• Speak naturally — the prompter follows you.
• Use *asterisks* for manual emphasis (works without AI).
• Blank lines create breathing spaces.
• Use Chrome or Edge for best support.
• For YouTube: aim for 140-170 WPM with frequent pauses.