I try to make the bot each time speaks it have a avatar speaking (picture open and close mouth)

Are you using any specific library for the visual part, I did something once but only passing the response to the google Text to Speech api. It seems to me that it would be the same implementation, but you must use the botpress api directly to receive the text and pass it to your library or platform that would take care of this, actually with respect to AI avatars I have only done it with platforms like veed.io. passing the text.