abstract |
Devices, systems and methods for recognizing speech in an audio signal are presented. In particular, various embodiments may indicate an end of one or more speech segments based at least in part on one or more characters predicted to be within those one or more speech segments. |