About Double Scribe
I built Double Scribe after one too many meetings where everyone agreed I could transcribe the call, but no one actually did it. I wasn't hosting the meeting, so the built-in transcribe button wasn't mine to press. Sometimes the other side was on a completely different platform that didn't have one at all. So Double Scribe runs locally on my own computer instead. It sits in the background, picking up my microphone and whatever's coming through my speakers, and I start and stop it myself, once everyone on the call has actually agreed to it.
Project goals
Double Scribe isn't trying to be a call-recording platform or a SaaS product. It's a small, offline tool for one job: turning a live conversation into a plain-text record you can search later, without an account, a subscription, or a server in the middle.
Under the hood
- Captures your microphone (Me) and your computer's audio output (Them) at the same time
- Transcribes locally with Whisper, on CPU or GPU: no internet needed after the first model download
- Splits speech into turns automatically and labels each one Me or Them
- Saves each conversation as a plain
.txtfile; the audio itself is never written to disk - Built with Python and a lightweight desktop UI, currently for Windows