Speaker recognition can recognise non selected speakers

* Ensure you have at least 2 speakers in the recognition system ( I used the evan dataset )
* Select the primary speaker in advanced backend
* Run some tests and check to see if evan is detected

Valid detection targets should only be the ones that are sent via the api request