Anti-Vocale

Transcribe voice messages offline. No internet required.

## Full Description (4000 chars max)

Too many voice notes and no time to listen? Turn them into text and read them whenever you want.

Anti-Vocale transcribes the voice messages of WhatsApp, Telegram, and any other app directly on your phone. It works on recorded calls and audio files too. Everything happens on your device: the audio never leaves your phone.

HOW IT WORKS
1. Long-press a voice message and share it with Anti-Vocale
2. Read the transcript that arrives in the notification
3. Copy it or send it back to the chat with one tap

PICK WHAT FITS YOU
- 99 languages out of the box (Whisper Turbo): the balanced choice for everyone
- Fast and light (Parakeet TDT, recommended): 25 European languages, two sizes (862MB high quality, 640MB compact)
- Text in real time (Nemotron 3.5): 45 languages, words appear while they are spoken
- One compact engine, 59 languages (Qwen3-ASR), including Hindi and Arabic
- Missing your language? Community models install in two taps (German, Spanish, Russian, Persian, Ukrainian, Hebrew and more), and you can import your own model file from a link or a folder: every file is integrity-checked on import

UNDERSTAND IT, NOT JUST READ IT
- AI summaries: long transcripts get a short, readable recap you can copy and share (needs a Gemma model; the prompt is customizable)
- Who said what: conversations get speaker labels (SPEAKER 1 / SPEAKER 2) in the transcript and in every export
- Any length works: long audio is split and stitched automatically, and if a run fails halfway it keeps the text it already produced
- Export as subtitles (SRT, VTT) or timestamped text

EVERYDAY COMFORTS
- Smart notifications: page through long transcripts without opening the app, one-tap copy or send back
- History with search: find any sentence again, retry with a different model
- Auto-copy and auto-save to a folder you pick (Drive, Syncthing)
- Video files too: the audio track is extracted and transcribed
- Faster by design: silence stripping, parallel processing, and automatic use of your phone's AI chip when available
- Per-app settings and Tasker automation

PRIVACY, FOR REAL
No account, no ads, no tracking, no data collection. After you download a model it works fully offline. Your audio stays on your phone, always.

OPEN SOURCE
Apache 2.0. Source code: https://github.com/RisorseArtificiali/anti-vocale

REQUIREMENTS
Android 8.0+ | 4GB+ RAM recommended | 326MB to 4.2GB of storage depending on the model