I'm not sure I'm right with the voice commands, but the /speak command supports very old versions of Monologue or Text Assistant. Agent speech recognition is supported by
Microsoft's simple SR software.
Hopefully someone will have more insight though, if I'm wrong
Regards,