Scrita turns speech into text: a voice note, a lecture, a meeting or an interview, with one speaker or up to six. Record or import audio, and Scrita tells the voices apart and writes a timestamped transcript with each participant's name.

PRIVATE
Your audio never leaves the phone. Internet is used once, to download the voice model (about 670 MB). No accounts, no ads and no analytics.

WHO SAID WHAT
Tells 1 to 6 voices apart and lets you name them.

25 LANGUAGES, ONE MODEL
NVIDIA Parakeet recognizes English, Spanish and 23 other European languages, with punctuation and capitalization, and detects the language on its own.

RECORD OR IMPORT
Record in the app, open a WAV, OPUS, OGG, M4A, MP3, AAC, FLAC, AMR or 3GP file, or share a voice note straight from WhatsApp or any other app.

RUNS IN THE BACKGROUND
Records and transcribes with the screen off, with progress shown in a notification.

EDIT THE RESULT
Fix a paragraph's text, change its speaker, join or delete paragraphs, or assign selected words to another speaker, with undo. Play back the original audio while you review.

EXPORT
Copy or share the .txt, or save it to Documents/Scrita together with the audio (M4A or WAV).

BACKUP
One .zip with all audio, transcripts and settings, to move to a new phone.

MAKE IT YOURS
Recording modes (clean voice, noise reduction, group, unprocessed), light or dark theme, and wallpaper colors (Material You).

FREE SOFTWARE
Open source under GPL-3.0: https://github.com/antruc/scrita

Requires Android 11 or later. An arm64 phone with 4 GB of RAM or more is recommended.