Scribe V2 is ElevenLabs' updated speech-to-text model, offering the highest accuracy with the lowest word error rates, advanced features like keyterm prompting and entity detection, and support for over 90 languages.

How does Scribe V2 improve on Scribe V1?

It provides better stability for long audio, improved handling of pauses/silences/tone changes, lower WER, automatic multi-language support, and new tools like keyterm prompting and entity detection.

What languages does Scribe V2 support?

Over 90 languages with automatic detection, including strong performance in diverse accents and multi-language files.

What is the pricing structure?

Free tier available; paid plans start low with included hours (e.g., around $0.40 per hour on average), and additional usage billed per hour depending on the tier. Keyterm prompting and entity detection incur extra costs.

Scribe V2

Most Accurate AI Transcription with Advanced Features

From ElevenLabs

Founder: Miko Matyska, Piotr Dabkowski

Audio & Music

9 Jan 2025

1 Million+

Freemium

Pricing Model

Freemium

Starting Price

$0/Month

About This AI

Scribe V2 by ElevenLabs is the latest speech-to-text model delivering the lowest word error rates on industry benchmarks.

It excels in batch transcription for long-form audio, with keyterm prompting, entity detection across 56 categories, smart multi-language support, speaker diarization, and word-level timestamps, ideal for subtitles, captions, and compliance workflows.

Audio Editor Text To Speech Transcript Voice Cloning Voice Design

Pricing

Pricing Model

Freemium

Starting Price

$0/Month

Key Features

Lowest WER on standard benchmarks for exceptional accuracy
Keyterm prompting for up to 100 custom terms
Entity detection with timestamps in 56 categories
Automatic multi-language detection and transcription
Smart speaker diarization and word-level timestamps

Pros

Superior handling of real-world audio challenges
Enterprise compliance and zero-retention options
Scalable API for developers and large workflows
Strong performance across 90+ languages
Integrated into ElevenLabs Studio for easy use

Cons

Higher tiers needed for high-volume transcription
Additional costs for keyterm and entity features
Web/API only with no native desktop app
Real-time mode (V2 Realtime) separate from batch V2

Scribe V2 is excellent for content creators, researchers, and enterprises needing precise, context-aware transcription with minimal errors and strong customization.

FAQs

What is Scribe V2?
Scribe V2 is ElevenLabs’ updated speech-to-text model, offering the highest accuracy with the lowest word error rates, advanced features like keyterm prompting and entity detection, and support for over 90 languages.
How does Scribe V2 improve on Scribe V1?
It provides better stability for long audio, improved handling of pauses/silences/tone changes, lower WER, automatic multi-language support, and new tools like keyterm prompting and entity detection.
What languages does Scribe V2 support?
Over 90 languages with automatic detection, including strong performance in diverse accents and multi-language files.
What is the pricing structure?
Free tier available; paid plans start low with included hours (e.g., around $0.40 per hour on average), and additional usage billed per hour depending on the tier. Keyterm prompting and entity detection incur extra costs.