Then the idea was to use all the information coming in as a constant feedback…
j previous speech k next speech
