Murmur: dictation that stays yours
Talking is faster than typing for a lot of us. Every cloud dictation service listens on someone else’s computer. Murmur sends your voice to a Whisper-compatible server you choose: one on your own hardware, or one we run.
What it is (and honestly, what it isn’t yet)
Murmur is a voice-to-text app being built by the person behind this site. It speaks the standard Whisper transcription API, so it works with any compatible server, including the free self-hosted one in our guide below.
It is in development, so there is no download link yet, and no screenshots we’d have to fake. What exists and works today is the dictation pipeline itself, running on my own hardware. The guide below is how you run the same thing.
Trying it yourself
The live demo isn’t running on this site at the moment. Dictation needs a speech server, and this site is currently published without one rather than routing your voice through anything we haven’t locked down. The build-log guide walks you through running the same thing yourself, which is the honest version of “try it” anyway.
Path 1: self-host it, free
One docker command on any reasonable PC (GPU optional). Our build-log guide is the exact setup this site uses, including the privacy rules that matter.
Run your own dictation serverPath 2: we host it, ~£3/month
For when “bind it to your LAN” is not how you want to spend a Saturday. Same software, running on our hardware, priced to cover the electricity. Beta waitlist only. The hosted service is not live yet and there is no billing. We’ll email you once, when it’s ready.