Did you manage to resolve this? I am getting the same MSG_pos errors, but I think it's because the current_speaker var is null.
Any chance you could do a tutorial video, or a write up of how to implement your system? The documentation doesn't explain a lot of details.