Large-Scale Speech Models

Alibaba's open-source large-scale speech model supports zero-shot cloning in 3 seconds, multilingual capabilities, and command-based emotional control, enabling ultra-low-latency streaming synthesis at 150 ms.