ShareSpeak Teleprompter: Getting Started and Troubleshooting

Do these steps once, on any computer:

1. Install and activate. Download ShareSpeak from the download page (on an Apple Silicon Mac take the Apple Silicon version, not Intel) and install it; the installation guide has the details. On first launch, click through the welcome screens to Ready to Go, paste the license key from your email and click Get Started (more in License key).
2. Choose your language and microphone. Open Customize → Language & Audio and check that Speech Language is the language you’ll speak: if it’s set to another one, the text won’t follow you (ShareSpeak 0.4.2 asks on first launch, pre-selected from your system languages; 0.4.1 and earlier start on English (US)). Under Microphone, System Default works; a headset or clip-on mic close to your mouth works better than a laptop mic.
3. Add a script. Write or paste it in the Scripts library (the main screen) and type a name in the title field above the editor (without one it shows as Untitled). It saves automatically.

Mac with Apple Silicon (M1 or newer)

1. In the dropdown under the editor, choose Apple Speech. It’s the default on Apple Silicon.
2. Press Play (Cmd+Enter). The prompter opens: set Speed, Font Size and Text Color here if you like.
3. Press Start in the header. The first time, macOS asks for access to the Microphone and to Speech Recognition: click Allow for both.
4. When the header says Listening..., start reading aloud.

On a MacBook with a notch, press Notch instead of Play to read from a panel right under the camera. Apple decides whether your speech is recognized on your Mac or on Apple’s servers, so keep the Mac online.

Mac with an Intel processor

Apple Speech, Local (Whisper) and the notch overlay aren’t available on Intel Macs (ShareSpeak 0.4.2 starts there in Classic mode and shows where to add a Gemini key). To have the text follow your voice, choose Gemini under the editor and paste your own Gemini API key (see Gemini). Without a key, choose Classic (scrolls at a speed you set) or Manual (you scroll), then press Play and Start.

Windows 10 and 11

1. Make sure Windows lets desktop apps use your microphone (see Permissions).
2. Under the editor, choose Local (Whisper, the default on Windows) and a model size next to it. Tiny is the fastest; Base and Small recognize other languages better but need a faster PC.
3. Press Play (Ctrl+Enter), then Start. The first Start downloads the model once: wait while the status shows Downloading Whisper model and Loading Whisper model, until it says Live. Once you start talking it shows Listening... (speech detected).
4. Read aloud. Local recognizes speech phrase by phrase, so the highlight moves in small jumps, usually a second or a few behind you. That’s normal.

Prefer the cloud, or speak a language Tiny handles poorly? Choose Gemini with your own API key instead. If Windows SmartScreen warns you when you first run the installer, click More info → Run anyway.

How to tell it’s working

• The timer runs and the status in the header says Listening... (Local: Live until you start talking, then Listening...).
• Words you’ve said turn green, and the line you’re on stays a little above the middle of the window.
• In the speech modes nothing scrolls on a timer: stop talking and the text waits for you.
• In Classic mode the status says End of script when the text reaches the end; in every mode, Rewind starts again from the top.

Don’t see the status or the AI mode switch? They’re hidden when the window is narrow, so make it a little wider.

• Play, then Start. Play (Cmd+Enter or Ctrl+Enter) opens the prompter with your script; Start in the header begins listening and the timer.
• Pause and Resume. Pause freezes the text and ignores what the microphone hears. The mic stays open, so Resume continues instantly, exactly where you stopped. Press Pause before you go off script and again when you’re back.
• Rewind (Cmd+[ or Ctrl+[) jumps back to the first word, while playing or paused.
• Double-click any word to continue from it: the words before it count as read and the text lines up on it. Use it whenever ShareSpeak loses your place.
• Scroll by hand at any time with the mouse wheel or trackpad, or Cmd+Shift+Up / Down (Ctrl+Shift+Up / Down). In Classic mode the auto-scroll carries on from wherever you scrolled to.
• Back or forward 5 words (0.4.2). In Apple Speech, Local and Gemini mode, Cmd+Shift+Up / Down (Ctrl+Shift on Windows) now move the reading position 5 words, as in the notch: the text follows your voice from there. In Classic and Manual they still scroll a third of the window.
• Full screen (0.4.2). The button in the prompter’s top-right corner, Ctrl+Cmd+F on a Mac or F11 on Windows (click the prompter first) shows the prompter alone on the screen its window is on; Esc or the Exit full screen button brings the window back.
• Sections, clicker and pedal (0.4.2). In a script with # headings, the Sections button or the [ and ] keys jump between them. A presentation clicker or USB foot pedal works once you turn it on in Customize → Clicker / Pedal, while the ShareSpeak window is focused.
• Stop ends the session and takes you back to the editor.
• How the prompter looks: set Speed (px/s), Font Size (16 to 120 px) and Text Color on the prompter screen before Start, and the window’s background in Customize → Interface Layout → Background Transparency. Line breaks and blank lines show exactly as you typed them. In 0.4.2, Display options below them add the font, line spacing, side margins, alignment, highlight colour, ALL CAPS, a reading line, a progress bar, a mirror for glass rigs, a countdown, a clock and a take summary; Reset brings back the defaults.

In the notch overlay (MacBooks with a notch): tap a word to jump to it, use Cmd+Shift+Up / Down to move 5 words back or forward, and press Esc to go back to the window. A line with only --- in your script starts a new notch page; long-press the next-page button to pick a page. Moving the mouse over the overlay pauses it (Pause on hover, in Customize → Notch Overlay).

Work down this list; most problems are one of the first three.

1. Check the reading mode. Under the editor, choose Apple Speech, Local or Gemini: Classic and Manual don’t listen. In the prompter header, the AI mode switch must be on.
2. Check the language. Customize → Language & Audio → Speech Language must be the language you’re speaking (see Speech language).
3. Read the status in the header. Permission denied means the microphone or speech-recognition access is off (see Permissions). Paused means press Resume. On Windows the status can say Live or Listening... even when Windows blocks the microphone, so check the privacy setting anyway.
4. Check the microphone. In Customize → Language & Audio → Microphone, try System Default, or pick the mic you’re actually speaking into.
5. Play doesn’t open the prompter? If the editor blinks red, the script is empty; if the API key field blinks red, Gemini needs your key.
6. It lost your place. ShareSpeak looks for your words a few words ahead of where you are, so skipping a whole sentence or improvising can leave it behind or push it ahead. Double-click the word you’re on (in the notch: tap it) and keep reading. Before an aside, press Pause.
7. The highlight stops during a long take while you’re clearly reading: press Stop, then Start (the text goes back to the top), then double-click the word you’re on. ShareSpeak 0.4.2 keeps Apple Speech going by itself: when Apple’s servers end recognition, about once a minute, it starts the next round at once (a few words at the switch can be missed; the highlight catches up), and if the Apple Speech helper fails it restarts it up to 3 times in 5 minutes before the header says Apple Speech stopped - press Stop, then Start.
8. Local trails behind you or the status says it’s running behind: see Whisper running behind.
9. Intel Mac: only Gemini follows your voice there (see Gemini).

macOS

The first time you press Start in Apple Speech mode, macOS asks for access to the Microphone and to Speech Recognition: click Allow for both. If you clicked Don’t Allow, or the status says Permission denied:

1. Open System Settings → Privacy & Security → Microphone and turn on ShareSpeak.
2. In System Settings → Privacy & Security → Speech Recognition, turn on ShareSpeak too.
3. Quit ShareSpeak and open it again.

In ShareSpeak 0.4.2 the header shows Permission denied with an Open System Settings button that opens the right page; turn ShareSpeak on there, then press Stop and Start.

ShareSpeak 0.4.2 never captures your screen, so it never asks for Screen Recording. Version 0.4.1 and earlier may show a Screen Recording prompt when you press Start; following your voice doesn’t depend on it, and you can turn it off in System Settings → Privacy & Security → Screen & System Audio Recording.

Windows

1. Open Settings → Privacy & security → Microphone.
2. Turn on Microphone access and Let desktop apps access your microphone (Windows 10: Settings → Privacy → Microphone → Allow desktop apps to access your microphone).
3. Restart ShareSpeak.

Open Customize → Language & Audio and choose the language you’ll speak under Speech Language before your first take. ShareSpeak 0.4.2 asks on first launch (“What language will you speak?”, pre-selected from your system languages); 0.4.1 and earlier start on English (US) whatever language you speak. There are 32 options: English (US, UK, Australia and India), Spanish (United States and Spain), French (France and Canada), German, Italian, Portuguese (Brazil), Dutch, Polish, Russian, Turkish, Arabic, Indonesian, Vietnamese, Hindi, Bengali, Gujarati, Kannada, Malayalam, Marathi, Tamil, Telugu, Japanese, Korean, Mandarin Chinese, Thai and Norwegian (Bokmål and Nynorsk).

• The Microphone picker sits in the same section.
• Gemini detects the language by itself.
• Local (Whisper) in a language other than English: Base or Small recognize it better than Tiny, if your computer keeps up. On Apple Silicon Macs, 0.4.2 adds Large v3 Turbo (574 MB), the most accurate outside English (see which speech model to use).
• Voice-following matches words separated by spaces, so scripts in Chinese, Japanese or Thai don’t track reliably; use Classic or Manual for them. Korean follows since 0.4.2.
• Apple Speech has no recognizer for Bengali, Gujarati, Kannada, Malayalam, Marathi, Tamil, Telugu and Norwegian (Nynorsk): choose Local (Whisper) or Gemini for those.

Gemini runs in Google’s cloud under your own API key, so your voice is sent to Google. It works on every Mac (including Intel) and on Windows, and it detects the language for you.

1. Create a key in Google AI Studio: follow how to get a Gemini API key.
2. In ShareSpeak, choose Gemini under the editor and paste the key into the Gemini API Key field (the “Get a key” link next to it opens the same guide).
3. Press Play, then Start. The status shows that the session is connected or listening.

Status mentions an invalid API key? Paste the key again without spaces, or create a new one in Google AI Studio.

Gemini stopped working? If the status shows an error or a closed session instead of listening, download the latest ShareSpeak from the download page and install it over your current version. Google retires older Gemini speech models, so an older ShareSpeak can stop connecting: 0.4.1 and earlier use a model Google shuts down by November 17, 2026, while 0.4.2 runs on Google’s current live transcription model, with nothing to set up. Meanwhile, the other reading modes keep working.

Status messages in 0.4.2: Listening... means connected. Gemini (sentence mode): Google’s backup speech model is in use, so the text moves after each sentence. Reconnecting…: the connection is being renewed (Google ends every session after about 10 minutes; ShareSpeak usually renews it in a pause). Gemini: invalid API key: check the key. Gemini unavailable – see Help: Google accepted none of its speech models for this key, so check that the key’s project can use the Gemini API. Gemini offline – Stop, then Start: no connection after several tries. The notch overlay can’t show these messages; it pauses instead, and resuming tries again.

Local (Whisper) recognizes speech phrase by phrase, so the highlight always moves in small jumps a second or a few behind you. If it trails much further, or the status says Whisper is running behind on this computer, your computer can’t run the chosen model in real time:

• Pick a smaller model under the editor: Tiny is the fastest. In 0.4.2 the status names the next size down (Large v3 Turbo → Small → Base → Tiny).
• Close heavy apps (video editors, games, browser tabs with video) while you record.
• On an Apple Silicon Mac, switch to Apple Speech.
• Or switch to Gemini with your own API key: the work happens in Google’s cloud.

Still lagging? Email us your diagnostics.log: in Local mode it records how long each recognition step took.

• Where the key is: in the email titled “Your ShareSpeak License Key”, sent from vlad@sharespeak.co after checkout. Check your spam or promotions folder.
• Where to paste it: ShareSpeak asks for it on first launch. Click through the welcome screens to Ready to Go, paste the key (Cmd+V on a Mac, Ctrl+V on Windows) and click Get Started. Activation needs an internet connection. The key isn’t entered in the settings, whatever older emails say.
• Copy only the key, with no spaces before or after it. It looks like XXXXXX-XXXXXX-XXXXXX-XXXXXX-XXXXXX-XX.
• Up to 2 computers per license.
• “License has reached activation limit” means the key is already active on 2 computers: reply to your license email.
• No email? Write to vlad@sharespeak.co from the address you paid with.

To update, download the latest version from the download page (or the Windows download) and install it over the old one. On a Mac, quit ShareSpeak, drag the new app to Applications and choose Replace. On Windows, run the new installer. Your scripts and settings stay.

ShareSpeak keeps a small support log. It records launches, which helper processes were running and, in Local mode, how long each recognition step took. It never contains what you said.

• Mac: ~/Library/Application Support/sharespeak-config/diagnostics.log. In Finder, choose Go → Go to Folder and paste ~/Library/Application Support/sharespeak-config.
• Windows: %APPDATA%\sharespeak-config\diagnostics.log. Paste %APPDATA%\sharespeak-config into the File Explorer address bar.

Attach it to an email to vlad@sharespeak.co with a line about what happened and when.

On Windows, ShareSpeak 0.4.0 and 0.4.1 can’t open their built-in instructions, and the Help & Shortcuts screen shows Instructions could not be loaded. The rest of the app works normally. Everything those instructions covered is on this page: start with Getting started.

ShareSpeak 0.4.2 fixes it: download it from the download page and install it over your current version (see Updating); your scripts and settings stay.

• My script shows up in a screen share or recording. Set Customize → Stealth Profile to Balanced or Ultra-Stealth; Visible lets capture see the window. The notch overlay has its own Hide from screen share switch in Customize → Notch Overlay. On Windows 10 before version 2004 the capture shows a black box where the window is (the text stays hidden). How the hiding works: invisible teleprompter for screen sharing.
• The window is invisible after switching to Visible. Restart ShareSpeak.
• ShareSpeak doesn’t open, or seems to be running twice. Only one copy runs; opening it again brings the first one forward. If you can’t see it, press Cmd+\ (Ctrl+\ on Windows) to show the window.
• The notch overlay doesn’t appear. It needs a MacBook with a notch (Apple Silicon, macOS 14 or later) with its built-in screen in use. On other Macs, read from the main window.
• Install problems (macOS won’t open the app, SmartScreen on Windows): see the installation guide.