A browser-based teleprompter with real-time voice synchronization. Your script scrolls automatically as you speak — no foot pedal, no remote, just your voice. It also generates timestamped subtitle files (SRT, VTT) as you record, ready to use in your video editor.
No install needed — open the app directly in your browser:
👉 https://gaufredev.github.io/syncprompt/syncprompt.html
Use Safari for the best experience (on-device speech recognition, works offline). Chrome and Edge work too but rely on cloud servers. Firefox probably won't work (voice sync requires the Web Speech API, which Firefox doesn't support).
If you want to run it locally, everything you need is already on your Mac.
Download this folder (green Code button → Download ZIP), then unzip it.
Open Terminal (search "Terminal" in Spotlight), then drag start.sh into the Terminal window. It will paste the file path for you. Press Enter.
First time only: if you get a "permission denied" error, type this in Terminal first, then try again:
chmod +x start.sh
The app opens automatically in your browser. That's it!
Why does this need Terminal? The microphone only works when the page is served from a local server (a browser security rule). The script starts a tiny server on your Mac — nothing is sent to the internet.
- Paste your script into the editor (left panel)
- Click Start and allow microphone access when your browser asks
- Speak naturally — the prompter follows your voice in real-time
- Use the speed and font size controls to adjust to your comfort
- Press E twice (or click End) to finish the take, then Export Captions for subtitles (SRT/VTT), YouTube chapters, or the take audio (turn on Record audio first)
During a take the script editor is locked, and closing the teleprompter window only pauses the take. Presentation clickers and foot pedals that send Page Down / Page Up move one line forward / back.
Flubbed a take, or want to skip ahead? Say a command instead of reaching for the keyboard:
| Command | Korean | French | English |
|---|---|---|---|
| Restart from the top | 처음부터 다시 해주세요 / 처음부터 다시 할게요 | reprends depuis le début | restart from the top |
| Restart this chapter | 챕터 다시 시작해주세요 / 챕터 다시 할게요 | reprends le chapitre | restart this chapter |
| Previous chapter | 이전 챕터 할게요 | chapitre précédent | previous chapter |
| Next chapter | 다음 챕터 할게요 | chapitre suivant | next chapter |
| Restart this line | 이 문장 다시 할게요 / 문장 다시 해주세요 | reprends la phrase | restart this line |
| Previous line | 이전 문장 할게요 | phrase précédente | previous line |
| Next line | 다음 문장 할게요 | phrase suivante | next line |
Spanish, German, Portuguese, Italian and Dutch have their own defaults: tick the language under Languages in the sidebar to see them under Voice commands.
Commands work in the language of the line you're on (not available for Japanese or Chinese). Change the phrases under Voice commands in the sidebar, and give a command several phrases by separating them with /. Start with a distinctive word: the first real word (not a tiny one like "이" or "on") must be heard clearly. If your script itself contains a command phrase, that phrase is switched off.
SyncPrompt automatically detects and switches between languages as you speak (Korean, English, French, and more). For best results:
- Select the languages used in your script from the sidebar
- Keep languages on separate lines — the system detects the language per line and adjusts voice recognition accordingly
- Mixing two languages on the same line works but is harder for the system to switch
---or## Chapter Title→ chapter breaks**bold text**→ emphasis- Lines starting with
@→ speaker mentions - Lines in parentheses like
(tone: serious)→ stage directions (ignored by voice matching)
| Problem | Fix |
|---|---|
| "permission denied" when running start.sh | Run chmod +x start.sh in Terminal first |
| Browser doesn't open | Go to http://localhost:8080/syncprompt.html manually |
| Mic not working | Click the lock/settings icon in the address bar → allow microphone |
| Voice sync feels off | Speak closer to the mic. On Mac, Safari tends to be more reliable than Chrome |
| "address already in use" | Another app uses port 8080. Edit PORT=8080 at the top of start.sh to 8081 |
- Safari (Mac) — Recommended. Voice recognition runs on-device (Apple Speech), works offline, most reliable.
- Chrome (Mac/Windows) — Works, but voice recognition uses Google servers (requires internet). May occasionally go silent during long sessions — the app detects this and auto-recovers.
- Edge (Windows) — Should work. Same engine as Chrome, not extensively tested.
Every website you visit uses energy — from the server that hosts it, through the network that delivers it, to the device that displays it. The average web page loads over 2 MB of data per visit, and the internet's total carbon footprint now rivals the airline industry's.
A lighter app doesn't just help the planet — it loads faster and runs smoother for you. SyncPrompt is built with this in mind:
- No Google Fonts — uses your device's built-in fonts instead of downloading ~300 KB of web fonts per visit
- No heavy frameworks — pure HTML, CSS, and vanilla JavaScript
- No tracking scripts — nothing phoning home in the background
- No unnecessary images — even the favicon is a lightweight inline SVG
Curious about the footprint of your own sites? Website Carbon Calculator · Green Web Foundation
MIT