# Speech Recognition

Alibaba Shrinks the Latency Gap: New Real-Time AI Speech Model Targets Global Livestreaming Dominance

Alibaba has upgraded its Fun-ASR-Realtime speech model, achieving near-offline accuracy with millisecond-level latency across 30 languages. The technology was recently validated during a 100-hour livestream, signaling a major push by the company to dominate the real-time AI transcription and translation market.