Note that none of the following applies to the SP version of VoxCommando which cannot be trained.
VoxCommando (regular version) uses the Windows desktop speech recognition as its engine.
By default the Windows speech recognition engine learns as you speak. This can be good as long as it is correctly understanding you and learning the right words. In general use it's probably a good idea to have this feature disabled or it will learn incorrect phrases and corrupt your speech profile to some extent. That is why it is turned off by default. You could turn it on when you are in a quiet room and are speaking clearly giving correct commands (phrases that exist) in order to train it to your voice but there are better ways to train the engine.
If you are in a multi user environment you might not want to train it all all.
But if you are the primary user and feel training is needed, then you are best off using the built in Windows speech training tools.
https://mcmw.abilitynet.org.uk/windows-10-speech-recognition-0Or if you have specific commands where you are having trouble getting them to be recognized then there is a more targeted tool in VoxCommando that you can use which is described here:
https://voxcommando.com/mediawiki/index.php?title=Profile_Training_Tool