Xiaomi unveils two AI audio frameworks: Any2Speech and Midasheng-audio-generate

小米AI语音新框架：人人都能当声音导演

Xiaomi's large-model application team introduced Xiaomi Any2Speech and Midasheng-audio-generate. Any2Speech generates up to about 10 minutes per inference, while the other model turns one text prompt into mixed audio with speech, music, and ambient sound. The post names GST labeling, dual-path planning with dimension dropout, Flow Matching, and five-field structured labels; benchmark scores, training scale, and commercial terms are not disclosed.

Why it matters: Xiaomi released two audio-generation frameworks with a clear hook and concrete mechanisms, so HKR-H and HKR-K pass. HKR-R is weaker because benchmark results, training data scale, open-source status, and commercial terms are not disclosed, so this sits at the low end of featured.