Well, it seems you took very little time reflecting.
My 2 cents:
The transscripts should be seperate events, that reference whatever event it is they are transscribing. This has a couple of benefits;
1: transscripts can be produced after the fact;
2: transscripts can be produced by anyone;
3: transscripts can apply to all sorts of events.
It would allow your client to provide transscripts for voice events that appearently did not have any.
This way your usecase can leverage stuff that is already out there, and other clients can ignore transscripts if they want to.
This set-up will benefit clients that would want to provide a good experience for deaf people for example, that can now addopt this broader transscript standard.
All a clients has to do, is also query for transscript events related to the stuff it wants transscripts for.
