BabbleScribe

Private voice typing

Say it out loud. Keep it to yourself.

BabbleScribe turns your speech into text in any app. Speech can be recognised on your phone, on your own server, or by a service you choose. Nothing goes anywhere you did not set up.

Coming soon to Google Play Get in touch

For Android 11 or newer on an ARM64 phone, for people aged 18 and over.

How your voice stays yours

Privacy is the starting point of the app, not a setting you hunt for.

On the phone first

Download a speech model, then dictate offline. Your audio never leaves the phone.

Your server, your key

Prefer more power? Point the app at a server you run, or add your own key for a cloud service. Audio goes only to the connections you add, and only for the task you route to them.

Encrypted on your phone

Notes and original audio stay on the phone, encrypted with a key held in the Android Keystore. A private session skips history and stats for one take.

No ads, no tracking

No account, no ads, no analytics, no crash reporting and no tracking libraries. BabbleScribe has no developer server, so the author never receives your audio, text or keys.

Open source, coming soon

The full source code will be public on GitHub soon, so anyone can check how BabbleScribe handles their voice. The app is coming soon to Google Play.

Speak in any app

Dictate where you already write, then keep what matters.

Voice keyboard

An optional keyboard with a microphone and no typing keys. Open it from the mic key of a keyboard such as HeliBoard or FUTO Keyboard.

Floating mic

An optional small microphone that appears over other apps while a keyboard is open, and inserts your words into the field you were using.

Polishing your words

Optional cleanup for punctuation, filler words and light grammar, using a connection you choose. It is off by default, and a failure always keeps your raw transcript.

Notes

Search, edit, pin, share and export your notes. Earlier versions are kept when the text changes.

Your own dictionary

Keep your own dictionary of names and words.

Insights on the phone

Weekly words, recording time, streaks and goals, worked out on the phone from your saved notes.

Many languages and scripts

Which languages you can dictate in depends on the speech model or service you pick. On the phone you can choose models such as Parakeet, Whisper, IndicConformer and Moonshine.

Your speech flow

Put your providers in the order you want. If one is unavailable, the app moves to the next one you added and tells you. It never uses one you did not add.

Encrypted backups

Backups go to a place you pick, with a recovery key that only you hold.

Where your words go

You decide, step by step. Nothing leaves the phone unless you add a connection.

  1. Always an option

    Your phone

    Speech is recognised by a model on the phone. Nothing is sent anywhere. Notes are stored here, encrypted.

  2. If you add one

    Your own server

    Point the app at a speech server you run. Audio goes only to the server you added.

  3. If you add one

    A cloud you chose, with your own key

    Add your own key for a service such as OpenAI, Groq, OpenRouter or ElevenLabs. Only the task you route to it is sent, and it handles that under its own terms.

One switch, Phone and my server only, keeps audio off every cloud provider.

Questions, answered plainly

Can I install it today?

Not yet. BabbleScribe is coming soon to Google Play. Questions are welcome at support@babblescribe.com.

Do I need an account or a server?

You need neither. There is no account, and you do not need a server. Download a speech model and dictate on the phone.

Does the author see my audio or notes?

No. BabbleScribe has no developer server, so the author never receives your audio, text or keys, and the app has no analytics or crash reporting.

What happens if I add a cloud service?

Audio and text leave your phone only for a speech or text service you set up, and only for the task you route to it. That service handles them under its own terms and privacy policy. There is no fallback to a provider you did not add.

Is it open source?

The full source code will be public on GitHub soon, so anyone can check how BabbleScribe handles their voice.

Which languages does it support?

Many. Which ones depend on the speech model or provider you pick, and BabbleScribe is multilingual by design.

Why does the floating mic need the Accessibility service?

The floating mic uses Android's Accessibility service to show the mic over other apps and to insert your words into the field you were using. It is optional, you turn it on yourself in Android Settings, and it never reads password fields. Everything else works without it. Read the privacy policy for the details.