Ted Hisokawa
Nov 11, 2025 09:20
ElevenLabs launches Scribe v2 Realtime, offering low-latency speech-to-text transcription in under 150 ms across multiple languages, enhancing live voice applications.
ElevenLabs has unveiled Scribe v2 Realtime, a cutting-edge speech-to-text model designed to deliver low-latency transcription for live applications. This latest development sets a new benchmark in the industry by providing real-time transcription in under 150 milliseconds across multiple languages, including English, French, German, Italian, Spanish, and Portuguese, as well as 90 additional languages, according to ElevenLabs.
Revolutionizing Live Transcription
Scribe v2 Realtime is tailored for applications such as voice agents, meeting assistants, and real-time captioning. It is engineered to handle complex scenarios and background noise, significantly outperforming existing models in the market. The model’s robust performance is attributed to features such as negative latency for next word and punctuation prediction, automatic language detection, and voice activity detection (VAD), which collectively enhance the transcription accuracy and efficiency.
Key Features and Compliance
The model offers a range of advanced features, including text conditioning, manual commit for transcript finalization, and support for various audio formats like PCM (48kHz) and μ-law encoding. Moreover, Scribe v2 Realtime is enterprise-ready, boasting compliance with major standards such as SOC 2, ISO 27001, PCI DSS L1, HIPAA, and GDPR. It also offers data residency options in the EU and India, along with a zero retention mode for sensitive workloads.
API Accessibility and Implementation
Developers can access Scribe v2 Realtime through the ElevenLabs API, enabling the integration of this powerful tool into various applications. The API allows for seamless deployment of natural, human-sounding agents, providing real-time understanding and response capabilities in live environments. This feature is particularly beneficial for developing voice assistants for customer support, sales, or in-product experiences.
Expanding Use Cases
With an impressive accuracy rate of 93.5% across 30 commonly used European and Asian languages, Scribe v2 Realtime is poised to transform how businesses and developers approach live transcription. The model’s ability to handle complex language scenarios makes it an invaluable asset for enterprises looking to enhance their customer interaction and operational efficiency.
For more information on Scribe v2 Realtime and to explore its capabilities, visit ElevenLabs’ official website.
Image source: Shutterstock
Source: https://blockchain.news/news/elevenlabs-introduces-scribe-v2-realtime
