Android · on-device dictation

On-device Whisper dictation for Android — genuinely offline

Q: Does OpenWispr's Android app work with no internet connection?

Yes. After the one-time model download, transcription and cleanup run entirely on-device. Put your phone in airplane mode and dictation still works — that's demonstrable, not a claim you have to take on faith.

Q: Is there really no free Android dictation app that's just as private?

Among the apps we could verify, few combine free, open-source, and confirmed on-device processing on Android specifically. Typeless, for instance, is closed source and cloud-only despite marketing that reads as local, and costs $12–30/month for its Pro tier. Where another app's Android architecture isn't publicly documented, we don't claim to know it — check the app's own source or documentation.

Q: How large is the model I have to download?

Currently a Whisper-class speech model, about 488 MB, downloaded once on first launch. From then on it runs fully on-device.

Q: Do I need a Google account or a separate OpenWispr account?

Install from Google Play like any app — no separate account or sign-in exists inside OpenWispr itself.

Q: Is my voice ever uploaded anywhere, even for improving the model?

No. OpenWispr has no servers to upload to. The app is MIT-licensed and open source, so this is a claim you can verify by reading the code, not one you have to trust.

Most "private" dictation apps still send your voice to a server. OpenWispr downloads a Whisper-class speech model once, then transcribes on your phone from then on — network or no network.

Download for macOS Get it on Google Play

Free on Google Play. No sign-up, no in-app account.

OpenWispr/Android/On-device dictation

Why this is rare on Android

On Android, "private" dictation usually still means "cloud, eventually"

Among Android dictation apps, few clearly combine free, open-source, and confirmed on-device processing in one product. Typeless, for example, ships a real Android keyboard with dictation, but multiple independent reviewers report it is cloud-only with no local mode — despite marketing copy ("Zero cloud data retention," "On-device history storage") that reads as local. It's also closed source and costs $12–30/month for its Pro tier after an 8,000-words/week free cap.

Other Android dictation apps exist whose on-device-vs-cloud architecture for the Android build specifically isn't documented anywhere we could verify — so this page won't name them as cloud-based when we can't confirm it, and it won't claim OpenWispr is the only genuinely on-device option, either. What we can say plainly: OpenWispr's Android app makes no network calls to transcribe, and the source is public so that's a checkable claim, not a marketing one.

How it actually works

One model download, then it's fully local

On first launch, OpenWispr downloads a Whisper-class speech model to your device (currently Whisper small, about 488 MB). After that one-time download, transcription and text cleanup both run on-device — put your phone in airplane mode and dictation keeps working exactly the same, because there's nowhere for it to fail over to a server.

This isn't a toggle you have to remember to turn on. There is no cloud path in the app to switch to — the architecture is local by construction, the same way it is on the macOS build.

Honest tradeoffs

What you're trading for that

On-device models are a real engineering tradeoff, not a free lunch: they're typically slower to start and less accurate at the extreme edges than a large cloud model with unlimited compute behind it. OpenWispr ships a right-sized model rather than the largest possible one, to keep the on-device experience usable on a phone rather than a data center.

OpenWispr's Android app is also younger than incumbents like Wispr Flow or Typeless — expect a smaller feature set today (see the comparison pages linked below for specifics on where OpenWispr is still catching up, like failure-recovery UX).