Live Interpreter
Not connected

Live Interpreter

One device captures. Everyone reads live bilingual subtitles.

I'm the host

Start a session on the phone or laptop that has the microphone receiver plugged in

  1. The speaker wears the Bluetooth microphone transmitter; plug the receiver into this device
  2. Set the receiver as the system input device (usually switches automatically when plugged in)
  3. Tap the button above, allow microphone access, and start recognition
  4. Share the QR code or link with your audience — they'll see bilingual subtitles

I'm in the audience

Enter the 6-character session code from the host

Or simply open the link the host shared — no code needed.

How it works

Speech recognition runs locally in the host's browser. Audio is never uploaded to any server — only the recognized text and its translation are broadcast to the audience over an encrypted channel.

For the host device we recommend Android Chrome or Chrome / Edge on Mac or Windows, where recognition is most reliable. Safari on iPhone also works, but keep the screen awake and don't switch to another app. Any mobile browser works for the audience.

Session ------

Share the link or QR code below with your audience

Languages

The audience sees the original and the translation together, and can switch to either one.

Microphone

Allow access first, then confirm the receiver is picking up sound

Live level —

Translation engine

The free engine needs no setup and works right away — fine as a quick fallback. An AI model gives smoother, more context-aware translations but needs an API key (stored only in this browser, never sent to the audience).

Use an AI model for real events. The free engine has a daily quota that a one-hour talk will likely exhaust, and it may be unreachable on some networks.
Only endpoints that allow direct browser calls will work. Many enterprise gateways (e.g. tokenhub) block requests from web pages for security — no way of filling this in will succeed; those require your own proxy. The ones below are verified to work.
Advanced settings

Defaults to the free EMQX public channel, which is good enough for most events. If the venue network blocks it or you need something private, point this at your own MQTT WSS endpoint.

Draft translation while speakingSends a draft translation before the sentence ends, for lower latency
Keep screen awakePrevents the screen from locking and interrupting recognition