ElevenLabs - Zero Runtime

ElevenLabs is a speech-to-text plugin. It uses ElevenLabs Scribe for real-time transcription, a good fit if you already use ElevenLabs voices and want broad language coverage.

Setup

Set your ElevenLabs API key in the worker environment. Generate a key from the ElevenLabs site:

export ELEVENLABS_API_KEY=<key>

Usage

Import the plugin and pass it to the pipeline’s stt slot.

from zeroruntime.plugins import ElevenLabsSTT

stt = ElevenLabsSTT()

# Pipeline(stt=stt, ...)

import { ElevenLabsSTT } from '@zeroruntime/js-sdk/plugins';

const stt = ElevenLabsSTT();

// Pipeline({ stt, ... })

Parameters

Constructor parameters for ElevenLabsSTT.

Parameter	Type	Default	Description
`api_key`	`str`	`None`	ElevenLabs API key. Falls back to the `ELEVENLABS_API_KEY` environment variable when unset.
`model_id`	`str`	`"scribe_v2_realtime"`	Scribe transcription model, e.g. “scribe_v2_realtime”.
`language_code`	`str`	`"en"`	Language of the speech as an ISO 639 code, e.g. “en” or “es”.
`sample_rate`	`int \| None`	`None`	Sample rate of incoming audio in Hz, e.g. 8000, 16000 or 48000. Leave unset to detect it automatically.
`commit_strategy`	`str`	`"vad"`	How finished transcripts are committed: “vad” ends a segment when the speaker pauses, “manual” waits for an explicit commit.
`vad_silence_threshold_secs`	`float`	`0.8`	Seconds of silence that mark the end of speech when committing by voice activity.
`vad_threshold`	`float`	`0.4`	Voice activity sensitivity, 0 to 1. Lower values pick up quieter speech.
`min_speech_duration_ms`	`int`	`50`	Shortest sound, in milliseconds, that counts as speech.
`min_silence_duration_ms`	`int`	`50`	Shortest pause, in milliseconds, that counts as silence.
`base_url`	`str`	`"wss://api.elevenlabs.io/v1/speech-to-text/realtime"`	WebSocket URL of the ElevenLabs realtime transcription endpoint. Change it only for a custom or regional endpoint.
`detect_language`	`bool`	`False`	Whether to detect the spoken language automatically and report it with transcripts.
`interim_results`	`bool`	`True`	Whether partial transcripts are also sent to the client alongside final ones. Turn detection uses partials either way.

Import paths

SDK	Import	Constructor
Python	`from zeroruntime.plugins import ElevenLabsSTT`	`ElevenLabsSTT()`
Node JS	`import { ElevenLabsSTT } from '@zeroruntime/js-sdk/plugins'`	`ElevenLabsSTT()`

Transcribed text passes to the LLM once turn detection decides the caller has finished speaking.