Speaker change on enrollments

Speaker change on enrollments is a feature based on Phonexia’s own diarization technology. It allows customers to be notified about the speaker changes during enrollment, which helps preventing a situations, where the enrolment is bad or compromised. Please note that this feature is currently available to single-server deployments only. To enable it, call endpoint /api/v2/maintenance/technologies in maintenance mode with the following JSON:

{
    "speaker_change_detection": "enable"
}

Once enabled, a voiceprint can be enrolled from a stream via endpoint /api/v2/enroll. Please note that the voiceprint created from a stream is checked for the number of speakers only after the stream has ended. If there is more than 1 speaker detected in the voice, error message is produced when using /api/v2/verify endpoint:

{
    "stream_uuid": "30f22400-2809-4ea3-8191-1f7289c3d009",
    "external_id": "JohnDoe",
    "detail": "Multiple speakers were detected in the voiceprint and it cannot be used."
}