Face & voice recognition
The FrameQuery desktop app can recognise people in your footage by face and by voice. This page explains what the app creates, where it is stored, what is uploaded for indexing and how to delete it.
Updated 17 October 2026
Off until you turn it on
Face recognition and voice recognition are optional and separate choices. You tick a consent box first.
Templates stay on your computer
Face and voice templates are created and stored on your computer. FrameQuery's servers never receive them.
Indexing uploads frames and audio
Frames can show faces and audio contains voices. They are not used to recognise anyone.
Face recognition and voice recognition are off until you turn them on. They are separate choices, so you can use one without the other. Before either runs, you tick a consent box.
That consent box records your own choice to use the features. It is not consent from the people who appear in your footage. See using FrameQuery for an organisation.
To produce captions, object tags and transcripts, FrameQuery uploads extracted frames and an audio track. It does not upload your original files.
Frames can show faces and audio contains voices. They are not used to recognise anyone. Speaker labels such as “Speaker 1” come from diarization of each file's audio in the cloud. No voice template or voice profile is stored.
FrameQuery's servers never receive face or voice templates, or the names you assign to people.
Analytics never include people's names, faces, file names or search text. The privacy policy describes what analytics do collect.
Face and voice templates and face thumbnails are encrypted on disk with AES-256-GCM. The key is held in your operating system's keychain or credential store.
The face matching index and people's names in the library database are not encrypted. They are protected by your computer's user account. Use your operating system's account controls and full-disk encryption to protect them.
Face and voice templates and the names you assign are never included in backups, shared indexes, exports, reviews or team (shared) libraries.
Face and voice recognition are not available in team libraries. Transcripts keep their speaker labels, such as “Speaker 1”, when you share them.
If you move your library folder to a network drive, the library database moves with it, including people's names. Face and voice templates and voice clips stay on your computer.
Uninstalling FrameQuery does not delete your local data. To remove face and voice data, use “Delete all biometric data” or withdraw consent in Settings before you uninstall.
When an organisation uses FrameQuery, it decides whether to use face and voice recognition. It is responsible for having a lawful basis for recognising people in its footage. This usually means the explicit consent of those people, and written consent where laws such as the Illinois Biometric Information Privacy Act (BIPA) apply.
FrameQuery does not receive the templates, so it does not process them on the organisation's behalf. Our Data Processing Addendum covers the frames and audio uploaded for cloud indexing.
You may not use FrameQuery's face or voice recognition for:
These restrictions are part of our End-User License Agreement.
The privacy policy, EULA, Data Processing Addendum and security page cover the rest of FrameQuery's data handling. For questions about face and voice data, email privacy@framequery.com.
