Dictate across apps
The floating bubble, voice-input panel, or Ascuta keyboard put words into the text field you are already using.
Private voice typing for Android
Turn your voice into words that stay yours.
Ascuta puts dictation into the text field you are already using. Tap the bubble, speak naturally, and keep moving.
Built around the useful part
Say less to your phone. Tap less, too.
Ascuta is a small, direct tool for getting speech into the app where you need it — and your words can stay on your phone while it works.
On-device first
Download a Whisper, Parakeet, Nemotron, or Gemma model once. Audio never leaves your phone, and no account is required.
Live words while you speak
Streaming engines show the transcript as you talk, then land the final text when you stop.
On-device AI post-processing
Clean up grammar, summarize, or translate locally — or with a cloud provider you choose.
Agent mode
Dictate instructions instead of words: Ascuta carries them out and inserts the result.
Optional cloud transcription
Use an OpenAI-compatible provider with your own key. Requests go straight from your phone to that provider.
Keep your keyboard
Ascuta complements the keyboard you already use instead of forcing you to replace it.
Up to 99 languages
Several multilingual models are supported — Whisper covers up to 99 languages with auto-detect.
Why Ascuta
How Ascuta compares.
In the Ascuta column, green is included free and blue is Lifetime Pro — so you can see how far the free app already goes.
| Feature | Ascuta | FUTO Voice Input | Wispr Flow |
|---|---|---|---|
| Pricing model | Free + one-time Pro | Free* | Free + subscription |
| Works fully offline | ✓ | ✓ | ✗ |
| Account required | None | None | Yes |
| Floating bubble in any app | ✓ | ✗ | ✗ |
| Keep your current keyboard | ✓ | ✓ | ✓ |
| Voice models to choose from | 4/14 | 3 | 1 |
| On-device AI post-processing | ✓ | ✗ | Cloud only |
| Agent mode: dictate instructions | ✓ | ✗ | Desktop only |
* FUTO offers an optional one-time payment to support development. Prices and features checked September 2026. See Pricing for what Free and Lifetime Pro include.
Questions, answered
Start simple.
The important details before you grant permissions or download a model.
Does local mode upload audio?
+
No. With a local engine, speech is processed on the device. Network access is used for model downloads and other operations you explicitly configure.
Which model should I download?
+
For most people, Whisper Base: small download, up to 99 languages, solid accuracy. Pick Parakeet for fast European-language dictation or Nemotron for live words while you speak. Models & hardware →
Which languages can I dictate in?
+
Whisper covers up to 99 languages; Parakeet v3 covers 25 European languages; Nemotron covers 32 locales; the system engine follows your Language setting. Supported languages →
Why does the bubble need Accessibility permission?
+
Android only lets an app insert text into another app’s focused field through Accessibility Service. Ascuta uses it for that insertion after you tap the bubble. Floating bubble →
Can I use Ascuta without the bubble?
+
Yes. Use the voice-input popup or select Ascuta as an on-screen voice typing keyboard. Ascuta Keyboard →
What does post-processing do?
+
It can clean up grammar and punctuation, summarize the transcript, apply spoken formatting commands, or follow your own custom prompt. Post-processing actions →
What is Agent mode?
+
Instead of dictating words, dictate an instruction — Ascuta carries it out and inserts the result. Agent mode →
What if an app does not accept inserted text?
+
Ascuta falls back to copying the transcript to the clipboard so it can be pasted manually. Troubleshooting →
Need help?
Setup should take minutes.
Use the docs for permissions, models, and common insertion problems.