Skip to main content

​
UltravoxRealtime

Ultravox realtime speech-to-speech (S2S) model.

​
Constructor

​
apiKey
string
​
config
UltravoxLiveConfig | null

​
cleanup

​
close

​
getRuntimeConfig

​
returns
Record<string, any>

​
UltravoxLiveConfig

Configuration for UltravoxRealtime.

​
Fields

​
model
string
Model id. Default 'fixie-ai/ultravox'.
​
voice
string | null
Voice id. Default null (sent as empty string).
​
languageHint
string
Language hint code. Default 'en'.
​
temperature
number | null
Sampling temperature. Default null (omitted).
​
maxDuration
string | null
Maximum session duration (e.g. '300s'). Default null.
​
timeExceededMessage
string | null
Message spoken when maxDuration is exceeded. Default null.
​
inputSampleRate
number
Sample rate (Hz) of the audio sent in. Default 48000.
​
outputSampleRate
number
Sample rate (Hz) of synthesized audio. Default 24000.
​
clientBufferSizeMs
number
Client-side audio buffer size in ms. Default 30000.
​
vadTurnEndpointDelayMs
number | null
Silence (ms) before a turn ends. Default 800; null omits the param.
​
vadMinimumTurnDurationMs
number | null
Minimum turn duration in ms. Default 600; null omits the param.
​
vadMinimumInterruptionDurationMs
number | null
Minimum interruption duration in ms. Default null.
​
vadFrameActivationThreshold
number | null
Per-frame VAD activation threshold in [0, 1]. Default 0.4; null omits the param.
​
firstSpeaker
string | null
Who speaks first, e.g. 'FIRST_SPEAKER_USER'. Default 'FIRST_SPEAKER_USER'.
​
enableGreetingPrompt
boolean
Have the agent open with a greeting prompt. Default false.
​
baseUrl
string | null
Override the API base URL. Default null.