ThousandQA New Speech Recognition Model Released: Medical Term Recall Rate Reaches 95.36%

According to Watchful AI monitoring, Alibaba has released the Qwen-Audio-3.0-ASR-Flash speech recognition model. It can understand speech in context, with a focus on improving the recognition of professional terms in fields such as medicine and industry.

Users can add custom hotwords to help the model correctly recognize drug names, equipment names, and industry abbreviations. The model can also correct spoken language expressions and automatically organize transcriptions with punctuation and structure.

In official internal testing, the medical term recall rate reached 95.36%, while the industrial term reached 93.24%. This series provides three interfaces for real-time streaming, audio file, and non-real-time recognition.