First month for $1Get your first month for only $1

Skip to content

Speech to Text API

Transcribe speech with ElevenLabs Scribe v3

Our most accurate Speech to Text model. Transcribe meetings, calls, and interviews, with speaker labels and controls for the words that matter.

Uh, hi! So, um, I was wondering if you wanted to meet up for coffee? Maybe tomorrow morning? [nervous laugh] Totally fine if not!

  • Lovable
  • Veed model
  • Synthesia
  • Stripe
  • Perplexity
  • Twilio

Most accurate Speech to Text API for batch workloads

Create captions, subtitles, and editable transcripts for podcasts, videos, interviews, and other recorded content – all with industry-leading accuracy via API.

Scribe v3Comparison modelWord error rate
20%0%5%10%15%20%
  • Scribe v2

    102 languages

    Scribe v3: 12.4%
    Comparison model: 18.5%
  • GPT-Transcribe

    58 languages

    Scribe v3: 5.5%
    Comparison model: 9.7%
  • MAI-Transcribe-2

    60 languages

    Scribe v3: 6.5%
    Comparison model: 7.5%
  • Muse Voice Transcribe

    25 languages

    Scribe v3: 6.6%
    Comparison model: 16.6%

Unprecedented transcription accuracy

Scribe v3 makes 27% fewer word errors than Scribe v2 across 102 languages, and fewer than every other batch model we tested.

Uh, hi! So, um, I was wondering if you wanted to meet up for coffee? Maybe tomorrow morning? [nervous laugh] Totally fine if not!

Designed for every scenario

Transcription that works in noisy environments, with background music, strong accents, and low-quality audio.

Fine-grained control over timing, speakers, and non-speech events.

The ElevenLabs Transcription API can detect laughter, emotion, and sound effects. Use keyterm prompting to guide transcription with domain-specific terms.

Transcribe audio and video

Upload MP3, MP4, WAV, MOV, and other common formats. Scribe handles files up to 15 hours with async processing and webhook notifications for large batches.
JavaScript code transcribing an interview video with ElevenLabs speech-to-text API.

Clean, editable transcripts

Get properly punctuated, paragraph-structured text ready for editing, publishing, or downstream processing. No cleanup required.
Editable transcripts

Keyterm prompting

Boost recognition accuracy for up to 1,000 domain-specific terms. Product names, technical jargon, and specialized vocabulary transcribed correctly the first time.
Keyterm Prompting

Dynamic audio tagging

Capture non-speech events like laughter, applause, music, and background noise. Transcripts include the full context of your audio, not just the words.

Smart speaker diarization

Automatically identify and label up to 32 speakers. Clear attribution of who said what, organized into readable transcripts.

Entity detection

Automatically identify and tag 56 entity types including names, dates, locations, and organizations within your transcripts.

Transcribe clinical audio with Scribe v2 Medical

A Speech to Text model optimized for medical transcription and clinical workflows, with enhanced recognition of medication names, anatomy, and pathology terms.

Medication renewal message flags metformin 500mg and asks about continued levothyroxine.

Transcribe clinical speech

Convert clinician-patient conversations, doctor dictation, and patient intake into precise text, ready for clinician review and use in notes, coding, and downstream documentation workflows.

Scribe v2 MedicalScribe v2 (Base)Word error rate
20%0%5%10%15%20%
  • Overall

    Scribe v2 Medical: 9.7%
    Scribe v2 (Base): 10.6%
  • In clinical sentences

    Scribe v2 Medical: 6.2%
    Scribe v2 (Base): 9.9%
  • Isolated drug names

    Scribe v2 Medical: 13.9%
    Scribe v2 (Base): 14.4%
  • Conversation

    Scribe v2 Medical: 6.1%
    Scribe v2 (Base): 6.1%

Fewer errors on medical terms

On term-WER, Scribe v2 Medical makes 15% fewer errors than Scribe v2 on the Eka Medical ASR test split, with improved medical voice recognition across clinical vocabulary.

Abstract gray textured background with soft diagonal light and dark bands.

Scribe v3

Highest accuracy, designed for batch workloads.

  • 27% fewer errors than Scribe v2
  • Lowest diarization error we tested
  • Number formatting and language hints
  • Entity detection and keyterm prompting
  • 90+ languages supported
Abstract grayscale bands of soft light and shadow sweeping diagonally.

Scribe v3 Live

Lowest latency, for realtime workloads.

  • Under 150ms Latency
  • 43% fewer errors than v2 Realtime
  • Keyterm library and multilingual mode
  • Improved turn detection
  • 90+ languages supported
Abstract grayscale texture with diagonal streaks and a soft gradient from dark to light.

Scribe v2 Medical

Medical fine-tune, for clinical transcription.

  • Medical Vocabulary Tuning
  • Transcription in 90+ Languages
  • Matches Scribe v2 on Everyday Speech
  • Improved Recognition of Medication Names
  • HIPAA-Eligible for Enterprise Customers

Transcribe speech in 90+ languages and a wide range of accents

Delivering exceptional accuracy across accents, dialects, and recording conditions.

Change the languageCode to preview languages

import { ElevenLabsClient } from "@elevenlabs/elevenlabs-js";

const elevenlabs = new ElevenLabsClient({
	apiKey: "<your_api_key>"
});
const response = await fetch(
  "https://storage.googleapis.com/eleven-public-cdn/audio/marketing/nicole.mp3"
);
const audioBlob = new Blob([await response.arrayBuffer()], { type: "audio/mp3" });

const transcription = await	elevenlabs
	.speechToText.convert({
	  file: audioBlob,
	  modelId: "scribe_v3",
	  tagAudioEvents: true,
	  languageCode: 
, // Set language diarize: true }); console.log(transcription);
Flag for en
English
Flag for zh
Chinese
Flag for es
Spanish
Flag for fr
French
Flag for pt
Portuguese
Flag for de
German
Flag for ja
Japanese
Flag for it
Italian
Flag for hi
Hindi
Flag for en
EnglishClick to preview

Powering the world’s leading companies and brands

  • “From dubbing Reels in local languages, to generating music and character voices in Horizon, ElevenLabs platform enables global creators, businesses, and enterprises to build with voice, music, and sound at scale.”
    Meta Color Logo
  • “Scribe’s unmatched accuracy across so many languages lets Fieldy understand every daily conversation and easily scale across continents. Fieldy has increased user retention by 50% after moving to ElevenLabs Scribe.”
    Fieldy logo
  • “ElevenLabs made it easy for us to quickly bring powerful text-to-speech capabilities to our SDK, allowing Agents to respond in real time with expressive voices to user questions or as feedback to what it’s seeing.”
    Stream Color Logo
  • “Twilio has integrated ElevenLabs’ generative AI voice technology into its CPaaS, enhancing ConversationRelay. This integration allows businesses and developers to create conversational AI voice interactions that sound human, feel expressive, and respond in real time directly from the Twilio CPaaS platform. We at ElevenLabs are excited that Twilio has chosen ElevenLabs to enhance ConversationRelay with the most expressive, human sounding voices available. ”
    Twilio logo

APIs built for production

Enterprise data protection developers

Data is encrypted in transit and at rest, with support for SOC 2, HIPAA, and GDPR compliance. EU Data Residency and Zero Retention modes are available for stricter data control.

Data is encrypted in transit and at rest, with support for SOC 2, HIPAA, and GDPR compliance. EU Data Residency and Zero Retention modes are available for stricter data control.

SDKs

Official Python and TypeScript SDKs with type safety, streaming support, and clear examples.

Our team ensures a smooth launch and keeps your operations running reliably, delivering consistent performance and peace of mind.

Frequently asked questions

Latest updates

    The most realistic audio AI platform