Send the room's sound. Captions follow.
A small app for the computer at the sound desk. Sign in with your InterScribe account; that's the whole setup.A small app for the computer at the sound desk. It carries your room's audio to InterScribe, so everyone gets live captions and translation in their language. Sign in with your InterScribe account; that's the whole setup.
The agent runs on a computer, not a phone. Open this page there to download.
Three ways to run it, one result
- 01Download and installPick the build for the desk computer. On a Mac, open the disk image and drag InterScribe Agent to Applications.
- 02Sign in and pick the channelUse your InterScribe account, choose the room's channel and the sound input, and allow the microphone.
- 03Test before it mattersIn InterScribe, press Test audio on the agent's card and speak. The meter moves when we can hear the room.
curl -fsSL https://staging.interscribe.io/install/agent.sh | shLinux ARM64/x64 and Raspberry Pi: installs a system service with sudo. macOS: installs a standalone binary after checksum and Apple notarization checks.
Linux: follow the installer’s account-specific configuration and status commands; the service starts automatically. macOS: use ./interscribe-agent login, then config and run. Keep that terminal open.
Already installed? Run the same command
Re-run this command to install the latest promoted release, including migration from a 0.x Linux service. Settings and the existing service account are preserved. Linux shows its plan and asks for confirmation before interrupting the service. On macOS, run it beside your existing binary and stop any foreground agent first; the signed updater does not restart it.
Afterwards, check the version, then status and doctor as the service account on Linux, or with ./interscribe-agent on macOS. Confirm the channel and microphone, then test audio in InterScribe.
Built-in updater (optional)
No sign-in or configuration is needed. The updater verifies the signed release and atomically replaces only the binary. It does not restart the agent.
Linux installed service updates
For a root-owned /usr/local/bin install, check without sudo and update with sudo. When audio can stop, restart the service explicitly. Replace SERVICE_ACCOUNT with the account printed by the installer; it may differ from your SSH login.
/usr/local/bin/interscribe-agent update --check
sudo /usr/local/bin/interscribe-agent update
sudo systemctl restart interscribe-agent@SERVICE_ACCOUNT.serviceStandalone binary updates
Stop a foreground agent when audio can pause. Update from its folder, then relaunch it yourself.
./interscribe-agent update --check
./interscribe-agent update
./interscribe-agent runFor an older Linux 0.x service without update, use the install command above. For an older macOS 0.x binary without update, stop the foreground agent, run the installer in a new empty folder, then deliberately replace the stopped old executable with the newly downloaded binary.
Install manually, every command
Paste this into a terminal, in a new empty folder you own. One line: it downloads the CLI, checks it against the checksum published with our release — and Apple notarization on macOS — then prints the version it installed.
No install command to show: we could not read a published CLI version. Reload this page to try again.
- 01Open agent.interscribe.ioOn the computer that hears the room, in Chrome, Edge or Safari. Nothing to install.Open the web agent ↗
- 02Sign in, pick the channelYour InterScribe account, the room's channel, and allow the microphone.
- 03Keep the tab openClosing it or letting the computer sleep stops the sound. Great for a one-off or a borrowed laptop.
Downloads
Get Windows from Microsoft Store, or choose a published direct installer for your computer.
Loading available versions...
- DownloadWindows (64-bit) · Microsoft StoreDesktop app · Windows 10 (19041+) or Windows 11 · Updates through Microsoft Store
Raspberry Pi and servers: use the command line above.
Which one fits your room?
| Desktop app | Command line | Browser | |
|---|---|---|---|
| Install | Once | One line | Nothing |
| Starts with the computer | Yes | As a service | No |
| Runs with nobody at it | Yes | Yes | Keep the tab open |
| Best for | The sound desk, every week | Raspberry Pi, racks, booths | A one-off or borrowed laptop |
Desktop app
Install once · starts with the computer · runs with nobody at it. Best for the sound desk, every week.
Command line
One line · runs as a service. Best for a Raspberry Pi, a rack, a booth.
Browser
Nothing to install · keep the tab open. Best for a one-off or a borrowed laptop.
Already streaming with OBS, vMix or an encoder for an online audience? An ingress takes that stream instead. It runs a few seconds behind the room, so it suits remote sessions, not people in the room.
Hardware that works
Mixers, USB interfaces, Raspberry Pi kits and fixed-room setups we have tested, with wiring diagrams.
Hardware guides ↗Questions
Do I need an agent key?
No. Sign in with your InterScribe account. Keys are only for a machine nobody signs into.
Which audio inputs work?
Anything the computer can hear as an input: a USB mixer or interface, a line-in, or the built-in microphone. You choose it in the agent when you pick the channel.
How do I update it?
The desktop app tells you when a new version is ready. On the command line, stop the agent, run ./interscribe-agent update, and start it again.