Got the speech demos working (eg 06.f.hybrid-speech) but there are at least two cases in which the activities sent from the bot are not spoken out loud:

a) proactive activities (like welcome message sent by the bot)

b) when user uploads a file

I understand that the bot only "speaks" when the user clicks in the mic button and talks in the first place.

But I need a way to "forward" utterances to be spoken "by the bot"

How can we control this behavior?
Is there some store event that we can dispatch?

Thanks in advance.