Search
On April 20, 2026, Tongyi Lab, a research entity under Alibaba, announced the launch of its cutting-edge large-scale speech recognition model, Fun-ASR 1.5. This versatile model boasts the capability to seamlessly handle 30 distinct languages, seven major dialects, and an impressive array of over 20 local accents. Notably, it excels in accurately transcribing recitations of ancient poetry, a testament to its sophisticated linguistic prowess. In typical dialect scenarios, the model has achieved a remarkable reduction in the character error rate by 56.2%, setting a new benchmark in speech recognition accuracy. Fun-ASR 1.5 is now readily accessible via Alibaba Cloud's BaiLian platform, offering robust API services to cater to a wide range of applications.
-
C114 Communication Network -
Communication Home
