agreed - we understand that speaker identification is crucial when it comes to extracting action items and we're actively working on this!