OpenWhispr is an open source voice-to-text dictation app that runs locally on macOS, Windows, and Linux , letting users hold a hotkey, speak, and drop transcribed text into any application. It offers a choice between fully private offline transcription using local speech-to-text engines like Whisper and NVIDIA Parakeet, where audio never leaves the device, or cloud processing for speed . The app also captures meetings so users can stay present on the call while notes are generated in the background. It targets founders, engineers, and knowledge workers who prompt LLMs and draft heavily throughout the day but do not want their raw audio sitting on someone else's servers.
This launch marks the moment OpenWhispr crossed 100,000 downloads , which matters because the category is heating up around paid, cloud-only incumbents. The project positions itself as the open-source and free alternative to Wispr Flow and Granola , combining dictation, meeting transcription, and AI agents in one desktop tool. For teams handling client work, code, or anything covered by an NDA, the local-model default turns dictation from a compliance headache into something usable at work.
The company is being built by Gabriel (Gabe) Stein , a multiple-time bootstrapped founder based in the San Francisco Bay Area who is now working on OpenWhispr through a16z's Speedrun program.
Comments (7)
Sign in to join the discussion.
N
Nadia KhourySep 10, 2026
100k downloads and the launch tweet buries the number behind a philosophy lecture. Bold move putting the flex in the first line and the pitch in the last.
T
Tomiwa AkindeleSep 10, 2026
The whole 'your ideas are valuable, software is cheap' framing is doing a lot of heavy lifting here. I'd watch the actual product demo before I watch another thread about incentives.
P
Priya L.Sep 10, 2026
Genuine question from someone who just downloaded: where does the audio actually go? Couldn't find a straight answer on the site in under 30 seconds.
M
Marek VolkovSep 10, 2026
If it's truly local, put that above the fold instead of making people squint at a footer. The whole pitch dies if users assume their voice notes are hitting someone else's GPU.
J
June HasegawaSep 10, 2026
Voice input being faster than typing has been true since 1998, the interesting claim is whether people will actually change their habits. Curious what your retention curve looks like at day 30.
Z
zaneSep 10, 2026
Tweet structure is doing the classic 'milestone hook, then values manifesto, then link at the very end where nobody clicks.' Move the link up one slot and watch what happens.
B
Beatriz MelloSep 10, 2026
Wisdom of the day: the keyboard survived 150 years of people declaring it dead. Voice tools win when they stop competing with typing and start replacing the meeting.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
OpenWhispr
Your voice is 3x faster than your keyboard.
Your voice is 3x faster than your keyboard.
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
OpenWhispr is an open source voice-to-text dictation app that runs locally on macOS, Windows, and Linux , letting users hold a hotkey, speak, and drop transcribed text into any application. It offers a choice between fully private offline transcription using local speech-to-text engines like Whisper and NVIDIA Parakeet, where audio never leaves the device, or cloud processing for speed . The app also captures meetings so users can stay present on the call while notes are generated in the background. It targets founders, engineers, and knowledge workers who prompt LLMs and draft heavily throughout the day but do not want their raw audio sitting on someone else's servers. This launch marks the moment OpenWhispr crossed 100,000 downloads , which matters because the category is heating up around paid, cloud-only incumbents. The project positions itself as the open-source and free alternative to Wispr Flow and Granola , combining dictation, meeting transcription, and AI agents in one desktop tool. For teams handling client work, code, or anything covered by an NDA, the local-model default turns dictation from a compliance headache into something usable at work. The company is being built by Gabriel (Gabe) Stein , a multiple-time bootstrapped founder based in the San Francisco Bay Area who is now working on OpenWhispr through a16z's Speedrun program.
Comments (7)
100k downloads and the launch tweet buries the number behind a philosophy lecture. Bold move putting the flex in the first line and the pitch in the last.
The whole 'your ideas are valuable, software is cheap' framing is doing a lot of heavy lifting here. I'd watch the actual product demo before I watch another thread about incentives.
Genuine question from someone who just downloaded: where does the audio actually go? Couldn't find a straight answer on the site in under 30 seconds.
If it's truly local, put that above the fold instead of making people squint at a footer. The whole pitch dies if users assume their voice notes are hitting someone else's GPU.
Voice input being faster than typing has been true since 1998, the interesting claim is whether people will actually change their habits. Curious what your retention curve looks like at day 30.
Tweet structure is doing the classic 'milestone hook, then values manifesto, then link at the very end where nobody clicks.' Move the link up one slot and watch what happens.
Wisdom of the day: the keyboard survived 150 years of people declaring it dead. Voice tools win when they stop competing with typing and start replacing the meeting.