Skip to main content

Usage guide

Setup, environment variables, and Python/JavaScript/Go usage examples.

MistralAISTT

Live speech-to-text with Mistral’s Voxtral realtime model, using your own Mistral key. Detects the language automatically, or can be pinned to one. Authenticates with MISTRALAI_API_KEY from your environment, or an explicit api_key.

Fields

str
default:"UNSET"
Voxtral realtime transcription model. Defaults to ‘voxtral-mini-transcribe-realtime-3-0’.
Optional[str]
default:"UNSET"
Two-letter language code, e.g. “en” or “ar”. Leave unset to detect the language automatically. Defaults to None (not set).
int
default:"UNSET"
How long the model may wait for context, 80 to 1200 or 2400 milliseconds. Lower is faster but less accurate. Defaults to 480.
Optional[list[str]]
default:"UNSET"
Names and terms to spell correctly, e.g. product or place names. Defaults to None (not set).
Optional[Mapping[str, Any]]
default:"UNSET"
Mistral’s own end-of-turn detection, e.g. {“min_silence_ms”: 300}. Use it when the pipeline has no VAD; leave unset to finalize on the VAD alone. Defaults to None (not set).
bool
default:"UNSET"
Emit partial transcripts while the user is still speaking. Defaults to True.
str
default:"UNSET"
WebSocket address of the realtime endpoint. Change it only when routing through a proxy. Defaults to ‘wss://api.mistral.ai/v1/audio/transcriptions/realtime’.