Stored recognised speakers for better transcription

As well as my voice, allow the other voices (linked to transcript segments) to be labelled/named and stored so that they can be recognised in future conversations.