There are a slew of use cases around taking in caption streams and systemically modifying them -- for instance, to split into sentences, to fix typos, to change language, to add additional metadata, etc.

Right now none of the current stream types really lend themselves to this kind of development. Lets consider adding a new kind of stream; name needs to be determined, but I'm thinking something like "Mutation" stream or "Processing" stream or "Programmatic" stream.

It is worth mentioning that there has not been much thought / documentation / support for running multiple streams through a single Opened Captions server. This kind of stream is likely going to become far more useful once that is in place (to be clear, even without that this kind of stream would be useful, it just might require the user to create a separate OC server per mutation which is not at all ideal.)