How to use a meeting transcription API: Real-time and batch with Scribe v2
ElevenLabs has launched Scribe v2 and Scribe v2 Realtime for automated meeting transcription. The service supports speaker-labeled, time-stamped outputs across 90+ languages.
Verified State Diff
Impact & Verification Analysis
Developers building meeting intelligence platforms, enterprise software architects, and companies requiring automated transcription workflows.
This expansion allows developers to consolidate their audio processing stack by using ElevenLabs for both transcription and voice generation, reducing dependency on third-party transcription providers like Deepgram or OpenAI's Whisper API.
Full Fact Overview
The release of Scribe v2 introduces a dual-mode transcription architecture supporting both batch processing and low-latency real-time streaming. By integrating speaker diarization and timestamping directly into the API, ElevenLabs is positioning its speech-to-text capabilities to compete with specialized transcription providers, moving beyond its core text-to-speech identity. The support for 90+ languages indicates a broad multilingual model deployment, likely leveraging their existing foundational audio models to improve transcription accuracy in diverse linguistic contexts.