Microphone audio
Speech received during a session.
- Used for
- On-device speech recognition.
- Processing and storage
- Microphone audio is used for real-time recognition on your computer and is not uploaded automatically.
Privacy and data
Speech recognition and script following happen on your computer.
Built-in recognition and following work offline. Scripts, terminology, and reading settings stay on your device and are not uploaded automatically.
Applies to TalkFollow 0.1.0 · Document updated: 2026-10-25
Three principles at a glance
On-device processing
Recognition and script matching do not rely on cloud services.
No account required
No TalkFollow account is required. Store acquisition may require a store account.
Network access when needed
Obtaining or updating the app and visiting the website require a connection.
How your voice guides the script
From hearing your voice to finding your place in the script, processing happens on your device.
Microphone input
Your selected microphone receives your voice.
On-device recognition
The built-in speech models delivered with the app recognize the content.
Script matching and following
Recognized content is matched to the script to update your reading position.
The app uses the selected microphone for recognition after you start a session. Opening a script alone does not start voice following.
What is processed or stored on your device
Local processing and local storage are different. Here is what each type of data is used for and what may remain on your device.
Recognition content
Text recognized by the on-device speech model.
- Used for
- Script matching and following.
- Processing and storage
- Recognized text is used for script matching on your computer and is not uploaded automatically.
Scripts and sessions
Imported scripts and saved sessions.
- Used for
- Reading scripts and reopening sessions.
- Processing and storage
- Script copies and saved sessions are stored locally. Original files stay unchanged.
Terminology
Term text, categories, and session terminology snapshots.
- Used for
- Terminology processing and script matching within a session.
- Processing and storage
- The terminology library and snapshots used by sessions are stored locally and are not uploaded automatically.
Speech models
The built-in primary and fixed auxiliary model files delivered with the app.
- Used for
- On-device speech recognition.
- Processing and storage
- Built-in models ship with the app and are read locally. Version 0.1.0 has no optional model installation or selection.
Reading settings
Display mode, font size, line height, and highlight colors.
- Used for
- Keeping your reading and floating-window display preferences.
- Processing and storage
- Settings are stored locally and are not uploaded automatically.
When a connection is needed
Local processing in the app and network access to the website are separate activities.
Using the TalkFollow app
- Recognition and following with built-in models work offline.
- Obtaining or updating the app through a store requires a connection. Store accounts and transactions are handled by that store.
- TalkFollow does not automatically upload speech, scripts, or terminology. Stores and system components have separate data-handling rules.
Visiting the TalkFollow website
- Browsing pages and accessing downloads require a connection.
- Website interface previews do not read local scripts, access your microphone, or save app settings.
- The on-device processing described here applies to the TalkFollow app. Website visits are network activity and may generate server access or error logs; this page does not promise a log-free website.
The Windows interface uses Microsoft Edge WebView2. Microsoft states that it collects required runtime information; optional information is controlled by Windows privacy settings. Component crashes may generate and send crash reports. The app includes Microsoft Defender SmartScreen, which collects and sends user information to Microsoft as described in the Microsoft Privacy Statement to help identify unsafe web content. This is separate from on-device speech recognition and script matching in TalkFollow.
Delete data and manage permissions
Sessions, terminology, and settings are stored on your device and managed by you. Protect your computer account, original scripts, and backups.
- Deleting a saved session from the session list removes its record and associated session artifacts. It does not delete your original script.
- Deleting an entry from Terminology does not change snapshots in saved sessions. Delete the corresponding sessions to remove their snapshots.
- You can revoke microphone access in your operating system privacy settings. Voice following cannot operate normally without microphone access.
Protect your content when reporting a problem
Start by describing the issue and how to reproduce it. Avoid sharing unnecessary personal information or script content.
Start with these details
- App version
- Operating system
- Steps to reproduce
- Error message
Avoid sending original scripts, recordings, or recognition content. Before sharing an error message, check it for personal information or script content.