Hand over text — the scenario runs a voicing program on your server and returns a finished audio file, with no outside services.