Privacy and Data

Privacy and Data

Voice2Win is designed around local processing. Normal microphone recordings, speech recognition, transcription history, and built-in artificial intelligence (AI) Enhancement run on your own computer.

What stays local by default

• Microphone audio.
• Speech recognition with downloaded Whisper, Parakeet, and stream models.
• Built-in AI Enhancement models and prompts.
• Transcription text, optional Waveform Audio File Format (WAV) recordings, settings, and logs.
• The local-network mobile remote connection when Local network mode is selected.

No sign-in is required for normal local use. Model downloads and application updates require internet access.

Features that can communicate outside the computer

Feature	Data path
Model download and update check	Connects to the relevant download or update source.
Mobile remote in local-network mode	Audio and commands remain on the selected local network.
Mobile remote with Tailscale Funnel	Traffic uses an encrypted internet route provided by Tailscale.
Conversation fact checking	The local AI processes conversation text; short encyclopedia search terms are sent to Wikipedia, but not the full conversation.
Ollama or LM Studio	Enhancement prompts are sent to the server address you configured.
Shared Voice2Win runtime	Audio transcription and/or enhancement workload data is sent to the configured Voice2Win computer.
Companion client	Microphone audio and shortcut actions travel over the same local network to the desktop host.

If an AI server or shared runtime is on another computer, “local AI” means local to that configured network environment, not necessarily the computer where you dictate.

Default data location

On Windows, the default application-data root is:

%LOCALAPPDATA%\Appshines\Voice2Win

This root contains settings and supporting files. Transcriptions default to its Transcriptions folder. Model storage also defaults under the application data area unless a different location is selected.

The companion client stores its independent settings on Windows under:

%LOCALAPPDATA%\Appshines\Voice2Win.Client\clientsettings.json

Moving or separating data

Under Settings → General → Storage Locations you can:

• Select a dedicated transcription folder.
• Select a dedicated main-model folder.
• Select an alternative folder specifically for AI Enhancement models.
• Move all application data to a folder below your user profile, for example .appshines\Voice2Win.

Moving the application-data root requires a restart. The move is performed early during startup so model workers and log files do not hold the data open. If individual items cannot be moved, Voice2Win reports them and retries on a later start.

History retention

• Saving WAV recordings enables playback and retranscription but increases storage use.
• Delete Existing WAV Files removes audio and preserves text.
• Delete old transcriptions automatically removes entries older than the configured number of days.
• Individual history entries and conversation passages can be deleted from their respective views.

Deletion can be permanent. Keep a backup before removing audio or relocating data if the history is important.

Network security guidance

• Keep mobile-device and companion-client services disabled when unused.
• Treat pairing and QR-code addresses as secrets because they contain access tokens.
• Prefer a trusted local network; use an optional password for shared runtime access.
• Install only the mobile CA certificate generated by your own computer and remove it when no longer needed.
• Keep Voice2Win versions aligned across shared-runtime computers.