Shruta: Encrypted Transcripts

Shruta records conversations on iPhone and Apple Watch, creates speaker-labeled transcripts on your device, and seals your content behind a passcode-derived key. There is no Shruta account or cloud backend for your content. FAIR PRICING • Try Shruta free: your first 7 recordings, with every feature included. • One purchase unlocks unlimited recordings. No subscription. No ads. RECORD, TRANSCRIBE, KNOW WHO SAID WHAT • Record conversations on your iPhone, or straight from your wrist with the Apple Watch app. • Transcription runs entirely on your device — in English, Dutch, and German, with regional English and German variants. • Speaker labels show who said what. Correct a label once and Shruta applies it across the conversation. If the automatic speaker count is off, reprocess with the exact number of participants — no re-transcription needed. • Optional Voice ID: name a voice once and Shruta can recognise it in later recordings. Opt-in, stored only on your device, deletable any time. SEALED, NOT JUST LOCKED • Your password or passcode is turned into an encryption key using Argon2id; content is sealed with AES-256-GCM. • No account or remote reset. Spookwerk has no backdoor and cannot recover your content. While an already-unlocked vault has its key in memory, device-owner authentication can authorise a local password change. • Face ID / Touch ID unlock is optional, for convenience. The app locks itself when you leave it. NOTHING LEAVES YOUR DEVICE • Audio and transcripts are processed on-device and never uploaded for processing. There is no content sync. • No analytics, trackers, advertising, or third-party analytics SDKs. Zero data collection. • Content leaves Shruta only when you explicitly export it: copy a transcript, or share it as PDF or text. The export menu marks this boundary clearly. APPLE WATCH • Start and stop recordings from your watch. Audio moves to your paired iPhone over WatchConnectivity and is sealed there. Requires your iPhone nearby with Shruta set up. Shruta (Sanskrit: “that which was heard”) is built in the Netherlands by Spookwerk for conversations that deserve to stay between the people who had them: medical appointments, difficult talks, interviews, coaching sessions, and everything you would rather not hand to a cloud. The first use of a speech feature may download on-device model assets. Once the required models are present, processing works offline.