VoiceAgentAzureSemanticVadTurnDetection interface
Azure semantic voice activity detection.
- Extends
Properties
| create_response | Whether a response is created automatically when speech stops. |
| end_of_utterance_detection | Semantic end-of-utterance detection configuration. Set to null to disable it. |
| idle_timeout_ms | Maximum idle time before the detector ends the turn, in milliseconds. |
| interrupt_response | Whether user speech may interrupt the agent's response. |
| languages | BCP-47 language codes used for speech detection. |
| prefix_padding_ms | Audio to include before detected speech, in milliseconds. |
| remove_filler_words | Whether filler words are removed from transcription. |
| silence_duration_ms | Silence required to end speech detection, in milliseconds. |
| speech_duration_ms | Minimum speech duration required to trigger detection, in milliseconds. |
| threshold | Activation threshold for voice activity detection, from 0 to 1. |
| type | The turn-detection type. Always |
Inherited Properties
| auto_truncate | Whether the input audio buffer is truncated automatically when speech stops. |
Property Details
create_response
Whether a response is created automatically when speech stops.
create_response?: boolean
Property Value
boolean
end_of_utterance_detection
Semantic end-of-utterance detection configuration. Set to null to disable it.
end_of_utterance_detection?: VoiceAgentEndOfUtteranceDetection
Property Value
idle_timeout_ms
Maximum idle time before the detector ends the turn, in milliseconds.
idle_timeout_ms?: number
Property Value
number
interrupt_response
Whether user speech may interrupt the agent's response.
interrupt_response?: boolean
Property Value
boolean
languages
BCP-47 language codes used for speech detection.
languages?: string[]
Property Value
string[]
prefix_padding_ms
Audio to include before detected speech, in milliseconds.
prefix_padding_ms?: number
Property Value
number
remove_filler_words
Whether filler words are removed from transcription.
remove_filler_words?: boolean
Property Value
boolean
silence_duration_ms
Silence required to end speech detection, in milliseconds.
silence_duration_ms?: number
Property Value
number
speech_duration_ms
Minimum speech duration required to trigger detection, in milliseconds.
speech_duration_ms?: number
Property Value
number
threshold
Activation threshold for voice activity detection, from 0 to 1.
threshold?: number
Property Value
number
type
The turn-detection type. Always azure_semantic_vad for Azure semantic voice activity
detection.
type: "azure_semantic_vad"
Property Value
"azure_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