OpenWhispr: Voice to Text
Download on the App Store — free, iPhone and iPad.
What you need
Setting it up
The app walks you through all of this on first launch. It’s written out here because the keyboard step trips people up, and because you can come back to it later if you skipped it.1
Install it and open it
Onboarding starts by itself. If you already pay for OpenWhispr, sign in
with the same account and Pro carries across — there’s nothing to buy
again.
2
Add the OpenWhispr keyboard
This is the part that makes dictation work in other apps. Onboarding takes
you into iOS Settings on OpenWhispr’s own page, where you:
- Tap Keyboards
- Enable OpenWhispr
- Turn on Allow Full Access
- Tap Allow on the popup
- Come back into the app
3
Allow the microphone
Nothing records without it. iOS asks once.
4
Pick Private mode or OpenWhispr Cloud
Private mode runs on your device and downloads a model the first time.
Cloud is faster and needs an account. You can change this per recording
afterwards, so the choice here isn’t binding. To use your own provider,
finish onboarding, then follow Bring your own key
below.
5
Switch to the keyboard in a real app
Open Messages or Mail, tap into a text field, then press and hold the
globe icon in the bottom-left corner of the keyboard and choose
OpenWhispr.
Why the keyboard asks for Full Access. An iOS keyboard extension can’t
reach the main app’s storage without it. OpenWhispr uses that shared storage
to pass your recording and the finished transcript between the keyboard and
the app — so without Full Access, the mic button in the keyboard has nowhere
to put the text and can’t work at all.
Dictating in another app
The flow is a little unusual the first time, and it’s deliberate rather than a bug:- Tap the mic in the OpenWhispr keyboard.
- iOS switches you into the OpenWhispr app, which starts recording, then switches you straight back to where you were. Recording keeps going in the background.
- Stop when you’re done. The transcript is pasted into the text field you started in.
What it does
On-device, OpenWhispr Cloud, or your own provider
You choose how your audio and text are processed in Account → AI Models. Each workflow has its own setting. On-Device (Private mode) transcribes entirely on your device. Audio never leaves the phone, it works with no signal, and transcription is unlimited and free — no account, no word count, no plan. It costs you a one-time model download and a little speed. OpenWhispr Cloud is faster and more accurate, needs an account, and is metered on the Free plan. Bring Your Own Key uses your own provider account. It needs no OpenWhispr account or subscription; your provider’s usage charges and limits apply. Audio or text for that workflow is sent to the provider you select, so this is separate from on-device processing. If you run out of cloud words, switching to Private mode removes the limit entirely. The app offers you that as the alternative to upgrading, and it’s a real one — see Cloud vs local processing.Bring your own key (BYOK)
Available on iPhone and iPad in app version 1.2.1 or later. If you don’t see Bring Your Own Key, update OpenWhispr from the App Store.- Open Account → AI Models in OpenWhispr.
- Choose a workflow: Dictation, Uploads, Text Cleanup, Note Formatting & Titles, or Chat & Voice Assistant.
- Select Bring Your Own Key, choose a Provider and Model, and paste your API Key. Get an API Key opens the provider’s key page.
- Tap Check Connection, then Save. Checking a connection doesn’t save your settings. Text checks send a short test prompt and may incur a small provider charge; transcription checks only check access to the model list.
- Repeat for each workflow you want to use with your provider. Setting up Dictation alone doesn’t configure cleanup, note formatting, or chat.
For Custom, enter your Server address and Model ID, plus an API key
if your server requires one. Use an address your phone can reach —
localhost
points to the phone itself. Discover Models is available for Custom and
OpenRouter; you can also enter a model ID manually.
Provider keys stay in secure storage on this device and aren’t synced. Workflows
using the same provider share its saved key; Custom keys are saved per server.
Live meeting recording doesn’t support personal providers yet. Use
On-Device or OpenWhispr Cloud for live meetings. In a managed workspace,
your organization’s settings may also restrict available providers.
What it costs
Downloading the app is free. So is on-device transcription, permanently.
Already paying? Sign in on the phone with the account you already have. Pro
is included and there’s nothing to buy again.
Managing an App Store subscription
If you subscribed inside the iOS app, Apple holds the billing, not us. That changes where you go:- Cancel or change it in iOS Settings → Apple Account → Subscriptions. We can’t cancel an App Store subscription for you — Apple doesn’t give us the ability, however much we’d like it.
- Refunds are Apple’s decision too, requested at reportaproblem.apple.com.
Android
Not yet — but it’s actively being built, and we’re really keen to get it to you. Plenty of people have asked, and we’re counting every one. Stay tuned for updates. We’ll say so the moment there’s something to install, and if you’d like us to tell you directly, email support@openwhispr.com — we’ll add you to the list, and we do write to everyone on it.Still need help? Email
support@openwhispr.com and tell us your
device, your iOS version, and the app version from the Version row on the
Account screen. The app’s own Send Feedback and Report a Bug buttons
reach the same inbox. A real person reads every message.Prefer to ask the community? Join us on
Discord.