On-device · MENA-aware
Transcription, translation, search, dialect detection — all running on your device. Your voice never leaves it.
Whisper-class quality, runs entirely on your phone or laptop. No audio ever uploaded for transcription.
Arabic → English, English → Arabic, dialect-preserving. Tap once on any voice message.
Ask "find the message about the contract" — your transcripts are searchable text the moment they're recorded.
MSA, Gulf, Egyptian, Levantine. The platform recognises which dialect you're speaking and adapts.
Mimi codec adapts from 1.2 kbps (satellite) to 6.0 kbps (WiFi). Voice always preserved.
Waveform · transcript · translation toggle · copy · save. Every voice message is a first-class document.
On-device STT runs alongside call audio. Captions appear in the language you choose.
Voice biometrics label who is talking on a multi-party call.
RNNoise / DeepFilterNet runs on outgoing audio before encoding.