Speaker recognition
Identifying which specific person a voice belongs to by comparing it with known voices. It isn’t the same as separating speakers.
Speaker recognition compares a voice with others already stored to say whose it is: “this voice is Ana’s”. To do that it needs a voiceprint for each known person.
- Diarisation: separates the people in a recording without knowing who they are.
- Speaker recognition: puts a name to a voice by comparing it with stored ones.
- Speech recognition (ASR): writes down what is said.
Because a voiceprint is used to identify a person, the GDPR treats it as biometric data, one of the special categories of data, with stricter rules.
In Echo, speaker recognition is optional and off by default. If you turn it on, Echo learns a voice when you name a speaker in the editor and recognises it in your later recordings; voiceprints are stored on the server and never sent to the browser. With it off, Echo only separates speakers and you add the names.
