ElevenLabs transcription

Set the ElevenLabs client's base URL to https://api.allmodels.io/el, then upload audio with speechToText.convert() or speech_to_text.convert().

Transcribe a file

import { File } from "node:buffer";
import { readFile } from "node:fs/promises";
import { ElevenLabsClient } from "@elevenlabs/elevenlabs-js";

const client = new ElevenLabsClient({
  apiKey: process.env.ALLMODELS_API_KEY,
  baseUrl: "https://api.allmodels.io/el"
});

const bytes = await readFile("call.wav");
const transcript = await client.speechToText.convert({
  file: new File([bytes], "call.wav", { type: "audio/wav" }),
  modelId: "scribe_v2",
  languageCode: "en",
  timestampsGranularity: "word"
});

console.log(transcript.text);

Transcription options

The response shape follows the ElevenLabs SDK types. Read transcript.text for the full transcript and inspect transcript.words when the selected model returns timestamps.

Select a provider

Submit provider preferences through the SDK request. JavaScript adds them to the request query; Python adds them to the multipart request body.

const transcript = await client.speechToText.convert({
  file,
  modelId: "scribe_v2",
  timestampsGranularity: "word"
}, {
  queryParams: {
    provider: { only: ["elevenlabs"] }
  }
});

The JavaScript SDK validates batch transcription model IDs before sending the request, so use an SDK-recognized ID such as scribe_v2 there. The Python SDK accepts canonical AllModels model names such as deepgram/nova-3.

For live microphones or calls, use the ElevenLabs realtime guide. See the file transcription API reference for every request field.