VoiceSdkAntispoofResult Struct Reference

Public Attributes
float	score
	Human score. More...
float	score_replay
	Replay attack score. More...
float	score_tts
	TTS (Text-To-Speech) attack score. More...
float	score_vc
	VC (Voice Conversion) attack score. More...

Member Data Documentation

score

float VoiceSdkAntispoofResult::score

Human score.

Scores closer to 1 mean it's a genuine human. Scores close to 0 mean it's a spoof.

score_replay

float VoiceSdkAntispoofResult::score_replay

Replay attack score.

Replay is an attack type when malicious agent records the genuine target voice and then plays the recording back to the speaker recignition device.

This value is reserved, it's equal to zero in default configuration. Specific antispoofing models/init data is required to retrieve this.

score_tts

float VoiceSdkAntispoofResult::score_tts

TTS (Text-To-Speech) attack score.

Replay is an attack type when malicious agent chooses the Text-to-Speech engine and then makes it pronounce target phrase to the speaker recognition device.

This value is reserved, it's equal to zero in default configuration. Specific antispoofing models/init data is required to retrieve this.

score_vc

float VoiceSdkAntispoofResult::score_vc

VC (Voice Conversion) attack score.

Voice Conversion is an attack type when malicious agent crafts a valid phrase from a set of genuine target user's recordings.

This value is reserved, it's equal to zero in default configuration. Specific antispoofing models/init data is required to retrieve this.