Transcribe

elevenLabs.transcribe requires exactly one of file or sourceUrl. modelId defaults to scribe_v2, word timestamps are enabled, audio-event tagging is enabled, and multichannel output defaults to separate transcripts. Optional inputs control language, diarization, speaker roles and count, channels, key terms, verbatim handling, temperature, seed, and file format. It returns one transcript or a transcripts array for separate channels.

automations/transcribe-audio.automation.ts

import { automation, markSignificant, onInvocation } from "automate.ax"
import { elevenLabs } from "automate.ax/elevenlabs"

export default automation("Transcribe audio", () => {
  const request = onInvocation<{ audio: File }>()
  markSignificant(elevenLabs.transcribe({ diarize: true, file: request.audio }))
})

Built-in triggers

Built-in actions

Runtime utilities

Integrations