Legal
Privacy Policy
What data Iskra handles, where your audio goes, and what stays on your device.
Last updated: May 28, 2026
Who we are
Iskra is a voice-typing app for macOS and Windows, built by an independent developer in Yerevan, Armenia. This policy explains what happens to your data when you use it.
Iskra is in beta. This policy may change as the product evolves; the date above reflects the current version.
Your audio and transcripts
When you dictate, the audio you record is sent from your device to our processing service and on to OpenRouter, which performs speech-to-text and the optional text polish. Iskra is not a fully on-device tool — your audio leaves your machine for the duration of that request.
We do not use your audio, transcripts, or dictionary to train models, and we do not sell them. Audio is processed to return your text and is not retained by us for any other purpose.
What stays on your device
Your transcript history and personal dictionary are stored locally on your machine in a SQLite database. Iskra does not upload that history back to us.
Audio retention on your device is your choice — keep nothing, the last day, or the most recent transcripts — and you can clear history at any time.
Credentials are stored in your operating system keychain (macOS Keychain or Windows Credential Manager), never in a plain settings file or log.
Device permissions
Iskra asks only for what it needs: the microphone to capture speech, Accessibility to paste text into the focused app, Input Monitoring for the hold-to-talk hotkey, and clipboard access to deliver the result. You grant these in your operating system settings and can revoke them at any time.
This website
This marketing site is static and sets no tracking or advertising cookies. See the Cookie Policy for details.
Contact
During the beta we do not operate a dedicated privacy contact address. We will publish one here before general availability.