OpenASR 0.1.0
The first public release of OpenASR: a local speech-to-text app for macOS, built on an Apache-2.0 open core. What works today, what doesn't yet, and where the code lives.
OpenASR 0.1.0 is the first public release. It is local speech-to-text: audio is transcribed on your own machine, with no accounts, no cloud transcription path, and no telemetry. The release has two parts — an Apache-2.0 open core that does all the work, and a macOS desktop app that wraps it in a product you can install and open.
This is a quiet first release, so the honest framing matters more than the pitch. Here is what it is, what actually works today, and what is not there yet.
What it is
The open core is the whole speech pipeline: audio capture, the local server, model verification, the fail-closed model pull chain, and the sandbox and path checks. It lives in the Apache-2.0 repository, and you can run transcription end to end from the command line without the app at all.
The macOS desktop app is a closed-source shell over that same core — onboarding, UI, packaging, code signing, and auto-update. It adds no hidden transcription path; everything that touches your audio is the open code you can read.
What works today
- File transcription with subtitle-grade output — drop in an audio file and get a transcript, including
srt/vttsubtitle exports. - Dictation, including real-time text insertion into other apps on the frame-sync streaming model.
- Live captions — speech transcribed as you speak, entirely on-device.
- Local models pulled from Hugging Face as
.oasrpacks, verified against a signed catalog before they are installed. - Local-only by default — no telemetry, no silent downloads, nothing leaves the machine unless you explicitly ask it to.
The app is macOS on Apple Silicon, signed and notarized, and it updates itself.
What’s not there yet
- Windows and Linux desktop apps. Both are planned; neither has a date. The open-source CLI already runs on Linux today.
- More model families. The current catalog is small and deliberately so — more languages, sizes, and accuracy/speed trade-offs will land as signed packs over time.
If either of those matters to you, the roadmap is the place to say so.
Where the code is
The core is on GitHub at QuintinShaw/openasr under Apache-2.0, with prebuilt CLI binaries on Releases. Read it, fork it, or run the pipeline yourself — the point of an open core is that you don’t have to take any of the above on faith.
The Mac app is a one-click download. That’s 0.1.0.