@yunsong Which language or voice are you using in your SSML? Can you cross-check from the speech studio audio content creation tool if the selected voice can use speaking style that you defined in this case?
If you are able to use the voice to express as or set the speaking style you can pickup the SSML from the studio and use them with your SDK too, viceversa.