Privacy
Nothing about the talk is stored.
The short version: your audio never reaches us, and no transcript or caption is ever written down. Here is exactly what does happen, checked against the code rather than the brochure.
01What happens
Where the audio goes
The microphone is read by your own browser, and the audio streams from that browser directly to the speech provider. It does not pass through a CrewEcho server, and there is nowhere in our system it could be recorded even if we wanted to record it.
The provider’s key never reaches your browser either. Our server mints a short-lived token — about ten minutes — and the browser uses that. That is the whole of our involvement in the audio path.
Where the text goes
Most target languages come straight back from the speech provider to your screens, on the same direct connection as the audio.
Some languages need a second hop, and that one does pass through us. Haitian Creole is the main example: the provider recognises your English, and the recognised English text is sent to a CrewEcho endpoint, which forwards it to a translation service and returns the Creole for your screen.
That text is in transit only. It is not written to a database, not written to a file, and not kept after the response is sent. But it does cross our server, and you should know that rather than read “nothing touches our servers” and assume otherwise.
What we store
- Your account. Company name, your email address, a hashed password, and which languages your organisation has enabled.
- A record that a session happened. Which organisation and user started it, the source and target languages, when it started, when it last checked in, when it ended — and the IP address it was started from.
- Demo requests. Whatever you typed into the request form, until we have dealt with it.
What we do not store
- Audio, in any form, for any length of time.
- Transcripts of what was said.
- The caption text that appeared on any screen.
- Anything at all about who was in the room.
The session record tells us a talk ran. It cannot tell us, or anyone who asks us, what was said in it.
Third parties
Speech recognition and translation are Microsoft Azure services. Your audio is subject to their handling of it, and the direct connection from your browser is to them. Hosting and the database are Railway. We do not use advertising trackers or analytics that follow you off this site.
Asking us to delete something
Since there is no transcript to delete, this comes down to your account and your session records. Ask and we will remove them. Deleting an organisation removes its users, its sessions and its history.