Five questions for any teleprompter that hears you
These are worth asking whichever product you choose.
- Where does speech recognition run: on your computer, or on a server? If it is a server, your voice travels, and with it most of your script.
- Where is the script stored, and is there an account? An account usually means a copy on someone else’s computer.
- What does the app connect to, and when? Ask for the list. Then watch.
- What happens to the audio: is it kept, or discarded as it is transcribed?
- What is left on the machine afterwards, and can you delete it?
Terera’s answers
- On your computer. No cloud speech service is involved. Audio becomes text on the machine and is discarded as it goes.
- Scripts, settings and rehearsal history are stored in your user profile on the computer. There is no account; you activate with a licence key.
- Two purposes. At setup, about 190 MB of speech model downloads from its public repository on Hugging Face; that host sees standard connection information, including your IP address, and nothing of your script or audio. For licensing, the app activates with the licensing service, Keygen, and re-checks periodically, using your licence key and a device identifier. A failed connection alone does not lock you out.
- Discarded as it is transcribed. After a run you get a short summary, stored on your machine. It is not a recording.
- On Windows, uninstalling Terera offers to remove its data. On a Mac, the data stays in ~/Library/Application Support/Terera until you delete it.
Watch it yourself
Install Terera and activate it while online. Then run a network monitor: on Windows, Resource Monitor (type resmon into Start) and its Network tab, which lists the processes with network activity and their connections; on a Mac, Activity Monitor’s Network tab, or nettop in Terminal. Use the app hard: paste a sensitive script, speak, go off script, open your run summary. Then disconnect from the network and confirm that reading and speech recognition carry on. For more depth, use Wireshark or your corporate proxy logs. If you see traffic that fits neither purpose, write to security@terera.app with the app version and what you saw.
Your computer also shows when something can hear you. Windows shows a microphone icon in the notification area of the taskbar while an app is using the microphone. On a Mac, an orange dot next to the Control Centre icon in the menu bar means the microphone is in use (a green dot, for a camera, takes its place when both are on), and Control Centre can show which apps are using it.
What this does not cover
- Machines that never touch the internet. Terera needs a connection for setup and for periodic licence checks. For a computer that may never go online, a manual scrolling prompter that needs no activation, or paper, is the right tool.
- Networks that block the download. If your network blocks huggingface.co, setup cannot fetch the speech model, and licensing needs api.keygen.sh. The model download starts at huggingface.co and can redirect to its file-delivery hosts. The security page has the detail for an allowlist.
- Sharing your screen. If you share your screen in a meeting, what matters is what the capture contains. On Windows, Hide from screen share asks Windows to leave Terera out of captures, and Microsoft says plainly that the mechanism is not a security guarantee. On a Mac, current macOS lets many sharing tools capture Terera anyway. Screen-share hiding depends on the capture method your app uses, so check your sharing preview before a live call.
- The room. A script on a screen can be read over your shoulder, and no software fixes a glass-walled meeting room.
- Formal attestations. If your organisation approves only software that carries a third-party security certification, write to security@terera.app before you buy. What Terera offers is a short list of network purposes you can check yourself, signed downloads and a named company.
For IT
Windows 10 and 11: a per-user install with no admin rights required, signed through Azure Trusted Signing, publisher TERKON PTY LTD. macOS 15.5 or later, Apple silicon or Intel: drag to Applications; signed with a Developer ID, TERKON PTY LTD, and notarized by Apple. The download links serve each download’s SHA-256 in a response header. Speech recognition is an on-device streaming recogniser (sherpa-onnx). Team and volume licensing: sales@terera.app. The full list is on the security page.
Terera is made by the team at Terkon Pty Ltd in Australia, who answer support@terera.app.
When something else is the better choice
- An air-gapped computer. See above.
- A team that wants scripts synced across everyone’s devices. A cloud prompter is built for that. Terera is built for the opposite.
- A script in a language other than English. Voice following works in English. Timed and manual scrolling work in any language.
Terera
USD 89 once, tax included. Three of your own machines, Windows or Mac. Twelve months of updates, and the installed version keeps working after that. A 30-day refund by email. No account.
Questions
Does Terera work without the internet?
Speech recognition runs on your own computer. Terera needs the internet for setup, when it downloads its speech model, and for periodic licence checks. A failed connection alone does not lock you out.
Does my script or my voice leave my computer?
No. Speech recognition runs on your computer, audio is discarded as it is transcribed, and scripts stay in your user profile. The app connects only to download its speech model and to manage your licence.
Do I need an account to use Terera?
No. You activate with a licence key that is emailed after purchase.
How can I check what Terera connects to?
Run a network monitor while you use the app: Resource Monitor on Windows, Activity Monitor or nettop on a Mac, or Wireshark or your proxy logs. Report anything that fits neither purpose to security@terera.app.
Can Terera be used on an air-gapped computer?
No. It needs a connection for setup and for periodic licence checks, so a manual scrolling prompter that needs no activation is the better tool there.
Sources
- Terera’s network behaviour, storage, install, signing and allowlist hosts: Security and the privacy policy. This page makes no claim those pages do not. Checked 1 October 2026.
- Wikipedia, “Resource Monitor”: launched as resmon.exe; the Network tab lists the processes with network activity and their TCP connections. Checked 1 October 2026.
- Apple Support, “View network activity in Activity Monitor on Mac”: the Network tab and the data your Mac sends and receives. Checked 1 October 2026.
- Microsoft Support, “Windows camera, microphone, and privacy”: a microphone icon appears in the notification area of the taskbar when an app is using the microphone. Checked 1 October 2026.
- Apple Support, “Use Control Centre on Mac”: the orange dot for the microphone, the green dot for a camera, one dot at a time, and the field that shows which apps are using them. Checked 1 October 2026.
- Microsoft Learn, “SetWindowDisplayAffinity function”: window content protection is not a security feature or a guarantee. Checked 1 October 2026.
- Electron documentation, BrowserWindow, setContentProtection: on macOS, apps that capture through ScreenCaptureKit capture the window anyway. Checked 1 October 2026.
Related guides: a teleprompter for Microsoft Teams, a teleprompter you pay for once and a teleprompter that follows your voice word by word. What it does in a meeting: Features. Where the speech recognition runs, compared: Terera vs Teleprompter.com and Terera vs Speakflow. All guides.