You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Apr 18, 2022. It is now read-only.
训练声学模型时，语言特征和声学特征是怎么实现一帧帧对应训练的？语言特征是按声韵母根据问题集得出特征，但是声韵母的时长是和声学特征时长不一样的，merlin中应该是5ms一帧，所以语言特征最后几维是不是加了相关特征？还望解惑！
非常感谢