eSpeak provides a hook for me to implement SSML's tag. I need to implement it in order to support CSS3 Speech's cue-before/cue-after.