VoiceAgentSemanticVadTurnDetection interface

Package:

@azure/ai-projects

OpenAI semantic VAD turn-detection settings.

Extends

VoiceAgentTurnDetectionConfig

Properties

create_response	Whether to generate a response automatically when the detector determines that the user has finished speaking.
eagerness	How quickly the detector concludes that the user has finished speaking. low allows longer pauses, while high responds sooner.
interrupt_response	Whether detected user speech automatically interrupts an ongoing response.
type	The turn-detection type. Always semantic_vad for semantic voice activity detection.

Inherited Properties

auto_truncate

Whether the input audio buffer is truncated automatically when speech stops.

Property Details

create_response

Whether to generate a response automatically when the detector determines that the user has finished speaking.

create_response?: boolean

Property Value

boolean

eagerness

How quickly the detector concludes that the user has finished speaking. low allows longer pauses, while high responds sooner.

eagerness?: "auto" | "low" | "medium" | "high"

Property Value

"auto" | "low" | "medium" | "high"

interrupt_response

Whether detected user speech automatically interrupts an ongoing response.

interrupt_response?: boolean

Property Value

boolean

type

The turn-detection type. Always semantic_vad for semantic voice activity detection.

type: "semantic_vad"

Property Value

"semantic_vad"

Inherited Property Details

auto_truncate

Whether the input audio buffer is truncated automatically when speech stops.

auto_truncate?: boolean

Property Value

boolean

Inherited From VoiceAgentTurnDetectionConfig.auto_truncate