Voiceprint
A numerical representation of a voice’s characteristics used to recognise the person. It counts as biometric data.
A voiceprint is a list of numbers summarising how a person’s voice sounds: timbre, pitch, way of speaking. Two recordings of the same person give similar voiceprints, which makes it possible to recognise them in new recordings.
When it’s used to identify someone, a voiceprint is biometric data and falls under the special categories in Art. 9 of the GDPR. An exception under that article is needed to process it, as well as clear information.
- Is it optional or always created?
- Where is it stored and who can use it?
- Can it be deleted?
In Echo, voiceprints are only created if you turn on speaker recognition for your account, which is off by default. They’re stored on Edorteam’s servers in the EU, never sent to the browser and only used to name voices you’ve already named. With it off, separating speakers stores no voiceprint.
