According to CoinMeta, Qianwen has released the qwen-audio-3.0-asr-flash speech recognition model, which combines contextual understanding to enhance the recognition of professional terms in fields such as medicine and industry. Users can add custom hotwords to help the model accurately identify drug names, device names, and industry abbreviations. The model can also correct spoken expressions and automatically organize the text into a formatted manuscript with punctuation. In official internal tests, the recall rate for medical terms reached 95.36%, and for industrial terms, it reached 93.24%. This series offers three interfaces: real-time streaming, recorded file recognition, and non-real-time recognition.