Vays

Vays is a FOREVER FREE voice notes app built around your voice and your privacy. Record a thought, a meeting, a task, or a conversation with multiple people. Vays transcribes directly on your device, using models tuned for Polish and English. Your audio is not sent to the cloud. Built for people who think out loud: • Construction and site managers capturing decisions in the field • Medical and clinical professionals, where accurate Polish transcription matters • Product managers turning meetings into tasks • Anyone whose best ideas come while walking, driving, or moving between meetings What’s inside • On-device transcription: your audio stays on your phone unless you decide otherwise • Sessions (BETA): multi-speaker recordings with speaker recognition. Your own voice is recognized from a one-time voice profile, and other speakers can be named with a single tap • Vays Dictation Keyboard: dictate in any app with on-device transcription • Local Gemma AI: chat with your notes offline, without an account • Polished transcripts: turn messy spoken language into clean, readable text • Voice tasks: dictate a task and Vays turns it into a title, notes, and category • Improve transcript: re-transcribe a memo using a larger Whisper model • Device sync: notes from your iPhone can appear on your Mac • No ads, no tracking, no account Privacy by default Vays is designed so your recordings stay yours. Recording, transcription, speaker separation, transcript polishing, and AI chat all run locally on your iPhone. There is no external AI service in the iOS data path. AI runs locally using Apple Neural Engine and the Gemma model. The only network features are optional iCloud sync, without audio recordings, and optional pairing with your own Mac running Vays on your local network. Optional companion: Vays for Mac Vays for Mac is a paid desktop companion for people who want richer analysis, support for long recordings, and system-wide dictation mode. Notes recorded on your iPhone can automatically appear on your Mac. Vays is built by one developer. Every piece of feedback is read, and every bug report is taken seriously.