
ElevenLabs has launched Scribe v2 Realtime, its most advanced speech-to-text system to date, capable of delivering live transcription with human-level precision in less than 150 milliseconds. The model supports over 90 languages, including 11 Indian languages such as Hindi, Tamil, Malayalam, and Telugu, and records a 93.5 percent accuracy score on the FLEURS benchmark—setting a new performance bar for real-time multilingual communication.
Tailored for developers and enterprises, Scribe v2 Realtime is built to enhance voice assistants, live captioning platforms, and meeting applications. It incorporates several technical breakthroughs, including negative latency prediction, which anticipates upcoming words to reduce delay; text conditioning, which maintains accuracy during long speech segments; and manual commit controls, allowing fine-tuned transcription management. Together, these innovations make it a robust solution for sectors such as customer service, healthcare, education, and compliance—where speed and reliability are critical.
To meet the specific requirements of the Indian market, ElevenLabs has introduced data residency features that enable organizations to store and process data locally in compliance with national regulations. The system also integrates seamlessly with ElevenLabs Agents, enabling developers to build conversational AI tools that deliver ultra-low latency, custom vocabulary support, and zero-retention privacy options.
Beyond speech technology, ElevenLabs has been expanding its ecosystem. The company recently launched Chat Mode, extending its capabilities beyond voice-only interaction, and entered the AI-generated music domain through partnerships with Merlin Network and Kobalt Music Group. These steps highlight its broader ambition to merge language, sound, and interaction under a single AI framework.
With Scribe v2 Realtime, ElevenLabs pushes the boundaries of live transcription—offering sub-second responsiveness, enterprise-grade accuracy, and extensive language coverage. The release reinforces the company’s commitment to making real-time, multilingual voice technology accessible and dependable across global industries.




