Ascuta: Offline Voice to Text

Ascuta is a private, offline voice-to-text app for Android. Speak instead of typing in any app and watch your words become text, transcribed on your device by open-source AI models — no account, no server, and no audio ever uploaded. It is free to start with no time limit, and a one-time Lifetime Pro purchase unlocks the larger, more accurate models and AI post-processing. A floating mic bubble sits on top of every app. Tap to start dictating, tap again to stop — or press and hold to talk. Your words are transcribed on-device and pasted straight into the field you're typing in, no copy-paste needed. Audio never leaves your phone in local mode: no account, no server, no upload. WHY ASCUTA • Truly offline speech-to-text — powered by open-source on-device AI models, so your voice stays private • Dictate from any app — messaging, email, notes, docs — via a floating bubble or the built-in voice keyboard • Live transcription — watch your words appear as you speak • AI post-processing — clean up grammar and filler words, or translate on the fly, with no API key required • Five transcription engines to choose from, so you can balance speed, accuracy, and language coverage • Multilingual — auto-detect across 99 languages, or pick your language for the most accurate results • Works as a keyboard — set Ascuta as a voice input method and dictate into any text field, just like Google Voice Typing TRANSCRIPTION ENGINES - ML Kit System (Gemini Nano / AICore) — streaming, no download required - Whisper — accurate, multilingual, three sizes to fit your device - Parakeet — fast, optimized for English and European languages - Gemma — the most accurate on-device model, ideal for long dictations - Optional OpenAI cloud mode — bring your own API key for cloud-based transcription and cleanup BUILT FOR PRIVACY Ascuta was built around one idea: your voice belongs to you. In local mode, everything happens on your phone — no audio is ever uploaded, no account is required, and no data is collected. Models are downloaded once, then run fully offline. SMART DICTATION FEATURES • Auto-stop on silence, so you don't have to tap to finish • Post-processing with on-device Gemma AI: fix grammar, remove filler words ("um", "uh"), or translate your dictation into another language — no cloud, no API key • Import your own compatible speech models for full control • Configurable bubble modes: toggle or hold-to-talk, whichever feels natural FREE TO START, UPGRADE IF YOU WANT MORE Ascuta is free to use — the floating bubble, the voice keyboard, and the Whisper Tiny/Base models never expire. A one-time "Lifetime Pro" purchase unlocks larger, more accurate models (including Gemma and Parakeet), live transcription, and AI-powered post-processing — no subscription, ever. Whether you're texting, taking notes, writing emails, or filling out forms, Ascuta lets you speak instead of type — quickly, accurately, and privately. Download Ascuta today and start dictating anywhere, offline. ACCESSIBILITY SERVICE Ascuta's floating bubble is optional and uses Android's AccessibilityService. When you enable it, the service detects the text field you have focused so Ascuta can insert or edit your dictation directly in it, and it draws the microphone bubble over other apps. If you turn on screen context, it also reads the text visible in the active window to improve dictation and Agent mode. Agent mode only writes the resulting text into the focused field; it never taps, navigates, or operates other apps. Everything is processed on your device — Ascuta never collects, stores, or shares this text — and you can turn the service off at any time in Android Settings → Accessibility. Ascuta is not an accessibility tool for disabilities; the service is used only for hands-free dictation.