ElevenLabsSTT
Real-time speech-to-text with ElevenLabs Scribe, using your own ElevenLabs key. A good fit if you already use ElevenLabs voices and want broad language coverage. Authenticates with ELEVENLABS_API_KEY from your environment, or an explicitapi_key.
Fields
str
default:"UNSET"
Scribe transcription model, e.g. “scribe_v2_realtime”. Defaults to ‘scribe_v2_realtime’.
str
default:"UNSET"
Language of the speech as an ISO 639 code, e.g. “en” or “es”. Defaults to ‘en’.
int | None
default:"UNSET"
Sample rate of incoming audio in Hz, e.g. 8000, 16000 or 48000. Leave unset to detect it automatically. Defaults to None (not set).
str
default:"UNSET"
How finished transcripts are committed: “vad” ends a segment when the speaker pauses, “manual” waits for an explicit commit. Defaults to ‘vad’.
float
default:"UNSET"
Seconds of silence that mark the end of speech when committing by voice activity. Defaults to 0.8.
float
default:"UNSET"
Voice activity sensitivity, 0 to 1. Lower values pick up quieter speech. Defaults to 0.4.
int
default:"UNSET"
Shortest sound, in milliseconds, that counts as speech. Defaults to 50.
int
default:"UNSET"
Shortest pause, in milliseconds, that counts as silence. Defaults to 50.
str
default:"UNSET"
WebSocket URL of the ElevenLabs realtime transcription endpoint. Change it only for a custom or regional endpoint. Defaults to ‘wss://api.elevenlabs.io/v1/speech-to-text/realtime’.
bool
default:"UNSET"
Whether to detect the spoken language automatically and report it with transcripts. Defaults to False.
bool
default:"True"
Whether partial transcripts are also sent to the client alongside final ones. Turn detection uses partials either way. Defaults to True.