Scribe v2 overview
Scribe v2 by ElevenLabs
Submit an audio file or URL to the speech-to-text endpoint and receive a transcript with optional speakers and timestamps.
Scribe v2 specs
- Model ID
elevenlabs/scribe_v2- Developer
- ElevenLabs
- Input
- Audio
- Output
- Text
- Endpoint
POST https://api.aimlapi.com/v1/stt/create
What Is Scribe v2?
Scribe v2 is a speech-to-text model from ElevenLabs for recorded audio. It transcribes speech in more than 90 languages and detects the language automatically, labels up to 32 different speakers, returns word-level timestamps and tags non-speech sounds such as (laughter) or (applause) in the transcript. It is available on AI/ML API as elevenlabs/scribe_v2.
Scribe v2 works on complete files. ElevenLabs offers a separate Scribe v2 Realtime model for live, streaming transcription; this page covers the batch model.
Specifications
- Available on AI/ML API since: 14 October 2026
- Input: audio file or audio URL, output: text transcript
- Languages: 90+, with automatic detection
- Speaker labels: up to 32 speakers
- Timestamps: word level by default, character level on request
- Audio events: tags such as (laughter) and (applause), on by default
Request Options on AI/ML API
A job is submitted to POST /v1/stt/create with either a url or an audio file, and the result is fetched from GET /v1/stt/{generation_id}. Optional parameters: language_code (ISO-639-1 or ISO-639-3) to skip detection, diarize to label speakers, num_speakers (1–32) as a hint for speaker labelling, timestamps_granularity (word or character), tag_audio_events, no_verbatim to drop filler words, false starts and non-speech sounds, plus temperature and seed.
Use cases
Meeting, interview and podcast transcription, where speaker labels for up to 32 people and word-level timestamps turn a recording into a readable, searchable transcript.
Subtitles and captions for recorded video, built from word or character timestamps.
Multilingual archives and call recordings, using automatic language detection across 90+ languages.
Clean transcripts for downstream text processing, with no_verbatim removing filler words and false starts, or audio event tags kept when laughter or applause matters to the content.
How to use the Scribe v2 API
View docs
Build with your agent
Give your agent the model details, code and documentation.
1Get your API key
Create API keyexport AIMLAPI_KEY="8f2b4c7d9a1e6b3f…"2Make your first request
curl 'https://api.aimlapi.com/v1/stt/create' \
-H "Authorization: Bearer $AIMLAPI_KEY" \
-H 'Content-Type: application/json' \
-d '{
"model": "elevenlabs/scribe_v2"
}'import os
import requests
headers = {
"Authorization": f"Bearer {os.environ['AIMLAPI_KEY']}",
"Content-Type": "application/json",
}
payload = {
"model": "elevenlabs/scribe_v2"
}
response = requests.post("https://api.aimlapi.com/v1/stt/create", headers=headers, json=payload, timeout=120)
response.raise_for_status()
print(response.json())const response = await fetch("https://api.aimlapi.com/v1/stt/create", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "elevenlabs/scribe_v2"
}),
});
if (!response.ok) throw new Error(`HTTP ${response.status}`);
console.log(await response.json());The request returns a generation id. Poll GET https://api.aimlapi.com/v1/stt/create?generation_id=<id> until the status is completed.
Models similar to Scribe v2
Scribe v2 FAQ
Is Scribe v2 available on AI/ML API?
Yes. Scribe v2 is live on AI/ML API as elevenlabs/scribe_v2. Submit a job to POST /v1/stt/create and fetch the result from GET /v1/stt/{generation_id}.
How many languages does Scribe v2 support?
More than 90, with automatic language detection. You can also set language_code to skip detection.
Can Scribe v2 tell speakers apart?
Yes. With diarize enabled it labels which speaker says each word, for up to 32 speakers. num_speakers sets the expected maximum.
Does Scribe v2 return timestamps?
Yes. Word-level timestamps are returned by default; set timestamps_granularity to character for character-level timings.
Is Scribe v2 the same as Scribe v2 Realtime?
No. Scribe v2 transcribes recorded audio files. Scribe v2 Realtime is ElevenLabs' separate model for live, streaming transcription.
How much does Scribe v2 cost?
$0.000079444 per second of audio, which is about $0.0048 per minute or $0.29 per hour.
What is the Scribe v2 model ID?
Use elevenlabs/scribe_v2 as the model parameter. Requests go to https://api.aimlapi.com/v1/stt/create with your AI/ML API key.