SpeakerVerifierOptions

Options & Configuration · AiDotNet.Audio.Speaker

Configuration options for the SpeakerVerifier.

Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → SpeakerRecognitionOptions → SpeakerVerifierOptions

For Beginners

The verifier answers two questions about a recording: “is this the same person as this other recording?” (verification) and “which of the enrolled people is this?” (identification). Each is a comparison score checked against a threshold, and the two thresholds are set separately here.

How It Works

#2090 note. This class existed but no constructor took it, so neither threshold had any effect: the model applied a single 0.6 to both decisions. Both now default to 0.6 — the value that was actually in force — and both are read. VerificationThreshold previously defaulted to 0.7 here, a number nothing ever applied.

Constructors

Constructor Summary
SpeakerVerifierOptions() Initializes a new instance carrying the model’s published defaults.
SpeakerVerifierOptions(SpeakerVerifierOptions other) Initializes a new instance by copying another instance.

Properties

Property Summary
OnnxModelOptions OnnxOptions { get; set; } Gets or sets the ONNX runtime options used when the embedding model is loaded from an ONNX file.
double IdentificationThreshold { get; set; } Gets or sets the similarity the best-matching enrolled speaker must reach before identification returns them rather than “unknown”.
double VerificationThreshold { get; set; } Gets or sets the similarity a pair must reach to be accepted as the same speaker.

Methods

Method Summary
void Validate() Throws if a value this model requires has been left unset or is not positive.