Interface STTOptions

interface STTOptions {
    apiKey?: string;
    enableDiarization: boolean;
    endpointing: number;
    interimResults: boolean;
    keyterm?: string[];
    language: string;
    sampleRate: number;
    smartTurn?: number;
    smartTurnTimeout?: number;
    vadThreshold?: number;
}

Properties

apiKey? enableDiarization endpointing interimResults keyterm? language sampleRate smartTurn? smartTurnTimeout? vadThreshold?

Properties

Optional apiKey

apiKey?: string

enableDiarization

enableDiarization: boolean

endpointing

endpointing: number

Silence duration in milliseconds before an utterance-final event is fired. xAI's default is 10ms, but we default to 100ms for better compatibility with LK EOT models.

interimResults

interimResults: boolean

Optional keyterm

keyterm?: string[]

Key terms/phrases (max 100, 50 chars each) to bias transcription accuracy toward.

language

language: string

sampleRate

sampleRate: number

Optional smartTurn

smartTurn?: number

Confidence threshold (0.0-1.0) for xAI's ML end-of-turn ("Smart Turn") prediction at silence boundaries. When unset, Smart Turn is left at xAI's default behavior.

Optional smartTurnTimeout

smartTurnTimeout?: number

Maximum silence in milliseconds (1-5000) before an utterance-final event is forced even when Smart Turn predicts the speaker is continuing.

Optional vadThreshold

vadThreshold?: number

Voice-activity detection threshold, range 0.0-1.0. Lower values capture quieter/noisier speech. When unset, xAI's websocket default (0.08) applies.

Settings

Member Visibility

Theme

On This Page

apiKeyenableDiarizationendpointinginterimResultskeytermlanguagesampleRatesmartTurnsmartTurnTimeoutvadThreshold