New Delhi: ElevenLabs, a voice and audio innovation powerhouse based in the United States, has introduced its most advanced low-latency Speech-to-Text (STT) model to date, Scribe v2 Realtime. With 93.5% accuracy, the new model provides human-quality transcription under one hundred and fifty milliseconds on the FLEURS benchmark with over 90 languages and 11 languages of India, including Hindi, Tamil, Bengali, and Telugu. It is designed to be used in real time to ensure that developers and enterprises can develop a seamless voice-based interaction in industries such as customer support, media, medical, and education.

Scribe v2 Realtime defines the new standards of the real-time multi-language communication of human understanding and immediate reaction. The model has negative latency predic

See Full Page