The Build Blog
OSWispa v0.4.1: a free open-source alternative to paid AI dictation tools
OSWispa is my local-first Whisper.cpp dictation app. Version 0.4.1 is the point where it starts feeling much closer to the “hold to talk, release to text” experience people expect from the paid AI dictation tools on the market, while staying open source and free for anyone to use.
If you want a polished subscription product, there are strong paid options like Superwhisper and Wispr Flow. If you want local control, source code you can inspect, and no monthly bill, OSWispa is the better answer.
The release that made OSWispa feel like a product.
The Linux path now keeps local Whisper contexts warm instead of rebuilding them on every dictation. That cuts the clunky re-init cost that was making short bursts feel slower than they should.
Wayland text insertion is much more stable, the app ignores synthetic keyboard events from `ydotoold`, and GPU use is more conservative so the desktop keeps headroom instead of getting pushed into unstable territory.
On macOS, there is now a Finder-friendly packaged app download, and first launch can auto-test the machine and pick a model that is realistic for the device instead of assuming every Mac should jump straight to a heavier setup.
Most voice tools want a subscription. This one is free for all.
No subscription wall
OSWispa is MIT-licensed and free to download, use, fork, and improve. There is no paid tier, no seat pricing, and no feature gate around the best local workflow.
Local-first by default
Your dictation stays on your machine unless you explicitly choose a remote backend. That matters if you care about privacy, latency, or just not routing every sentence through someone else’s SaaS.
Built for real hardware
The new first-run flow tests the device it is running on and picks a model that aims to stay responsive. Older Intel Macs usually land on `base.en`. Bigger Apple Silicon and GPU boxes can stretch further.
A fair comparison, not pretend objectivity.
Paid AI dictation apps like Superwhisper and Wispr Flow are more polished out of the box. They put a lot of work into onboarding, UI, and the kind of smoothness people expect from a commercial desktop app.
OSWispa is solving a different problem. It is for people who want a local, hackable, open-source dictation tool they can actually inspect and change. If your priority is convenience and you do not mind paying, the paid tools are worth a look. If your priority is control, privacy, and zero subscription cost, OSWispa is the better fit.
The point is not to pretend open source automatically wins. The point is that a free, local-first option should exist at all, and now it is good enough that more people can use it instead of defaulting straight to monthly SaaS.
Everything in one place.
Who OSWispa is for
If you want a free open-source dictation app for Mac or Linux, and you care more about local control than polished SaaS onboarding, try it. If you want the smoothest commercial UX and you are happy to pay for it, the paid tools still exist. It does not have to be one religion or the other.