Privacy Policy
LAST UPDATED: 29 SEPTEMBER 2026
Voice Floww is a voice dictation tool with microphone and keystroke access — a level of system access you should never have to take on trust. This policy explains, in plain language, exactly what data moves where. It is written to match the product as built, and its core claim is checkable: Windows' own microphone-usage indicator lights up only while you hold the hotkey.
1. There is no Voice Floww server
Voice Floww has no accounts, no sign-in, no cloud sync, no billing and no usage tracking. We operate no server that receives anything from the app. The only network traffic the app generates is described in section 2: your dictation goes to providers you have chosen and hold your own API key with, and crash reports, while you leave them on, go to Sentry.
2. What leaves your computer
| Data | Where it goes |
|---|---|
| Your voice audio | Held in memory while you hold the hotkey, then sent to Groq's transcription API (api.groq.com) with your own API key and discarded. If you add an OpenRouter key and speak Hindi, Bengali, Spanish, French, Chinese or Japanese, it goes instead to Fish Audio's transcription model through OpenRouter (openrouter.ai), with your OpenRouter key; if that fails, to Groq. It is never written to disk by Voice Floww. |
| The transcribed text | Sent to a Groq language model for clean-up (or, only if you configure an OpenRouter key and Groq is unavailable, to OpenRouter), then pasted into the application you are typing in. |
| Window title and app name | Included in those requests so the formatter can match the context (e-mail vs. code). Only the title of the window you are dictating into — never screenshots, never other windows. |
| Clipboard content | Off by default. If you switch on clipboard context, the current clipboard text is included in the formatting request. |
| Crash reports | Only when the app hits an error, and only while Send crash reports is ticked in the tray menu (it is on by default; untick it to stop them). A report holds the kind of error, which part of Voice Floww's code it happened in, whether the app recovered, the app version and the Windows version. It goes to Sentry (sentry.io) and is stored in the EU. It never holds your audio, your text, the error's message, your computer's name, your Windows user name or your file paths. Sentry sees your IP address as part of the connection, as any website does; it is set to store neither the address nor the location it points to. |
| Version check | Once per launch the app fetches a small file from voicefloww.com to learn whether a newer version exists. No identifier of you or your machine is sent. |
Groq's and OpenRouter's handling of that data is governed by their own privacy terms, under the account you created with them.
3. What stays on your computer
Everything else lives in %USERPROFILE%\.voicefloww: your API keys, your dictionary and snippets, an optional local history of your dictations, and the app log. Delete that folder and it is gone. Nothing in it is ever uploaded.
4. Your rights under the DPDP Act, 2023
Because Voice Floww collects no personal data (crash reports are built to hold none), there is nothing for us to access, correct or erase on your behalf; the data is on your own device. For anything sent to Groq or OpenRouter, exercise your rights with them directly. If you believe this policy is inaccurate, write to [GRIEVANCE_EMAIL].
5. Security
Data in transit is encrypted (TLS). The microphone is opened only while the hotkey is held; the Windows microphone indicator shows this. You can confirm the app's network destinations with any network monitor or firewall log.
6. Changes to this policy
We'll update the "last updated" date above whenever this policy changes materially.
7. Contact
Voice Floww is published by [LEGAL ENTITY NAME]. For any privacy question, write to [GRIEVANCE_EMAIL] or support@voicefloww.com.