use language parameter for multiligual output where necessary

Some work to do. How to guess a film title or footballer's name should be spoken in English?
At the moment, the default way is the best we've got so far.

Multilingual output is necessary when it comes to reading song titles from discographic notes.

No file, test, or entry point is named. Start by locating the code that produces spoken output for film titles, footballers' names, and song titles, then define how a language parameter should be selected and verify that multilingual titles are pronounced appropriately.

Written by the indexing model from the issue text.