Privacy by design

Speak Flow Privacy Policy

Speak Flow turns your voice into text through a secure proxy to Sarvam AI, with optional chat powered by Google Gemini. This policy explains exactly what audio and data we touch, what we keep, and what stays on your device (2026).

What Speak Flow collects

Speak Flow is a voice dictation tool. We only collect what is needed to turn your speech into text and keep your account working.

  • Audio you actively dictate: raw PCM audio is streamed to our server only while you are holding the bubble or pressing record, then discarded once the transcript is returned.
  • Transcripts and edits you choose to save to your history, personal dictionary, or snippets.
  • Account details from either Google Sign-In (name, email, avatar) or email/password registration. Passwords are never stored in plain text — they are hashed server-side before being saved.
  • Basic usage diagnostics — session timestamps, word counts, and error reports — to keep dictation accurate and reliable.

How your voice is processed

Speech-to-text happens through a privacy-conscious proxy pipeline. Your audio is never stored long-term.

  • Audio is streamed over a secure WebSocket (wss://) from the app or browser to our FastAPI proxy.
  • Our server forwards the stream to Sarvam AI for transcription and returns only the text back to your device.
  • We do not retain raw audio after a dictation session ends. Only the final transcript (if you save it) is kept.
  • The Chat feature uses Google Gemini to process text you type or dictate into the chat screen.

What we store on your device

Most of your personal data lives locally, not on our servers.

  • History, personal dictionary, and snippets are stored in on-device storage (Hive on mobile, browser storage on web).
  • Your server endpoint, preferred language, and auth token are kept in local preferences for quick sign-in.
  • You can clear any of this at any time from the Settings screen without contacting us.

Retention, security, and your rights

We keep account data only while you use Speak Flow, and protect it with standard safeguards.

  • All traffic between the app and our server uses TLS. Credentials are hashed and access is limited to the minimum team needed.
  • Deleting your account from Settings → Danger Zone removes your profile, saved transcripts, and credit history from our database.
  • You can request an export or full erasure of your data at any time by emailing us.
  • We do not sell your voice, transcripts, or personal data to third parties.

Contact & legal

If you have questions about this policy, want to exercise your privacy rights, or need help with your account, email us at npmtechsolutions@gmail.com. We will respond within 5 business days.

Speak Flow is operated by NPM Tech. We reserve the right to update this policy; last updated April 2026.