I import and generate audio for 3 articles per day.
Overtime, it gets boring earing always the same voice, so I change voices.
Problem: it requires diving in the global configuration, and is slow, while it should be a quick local (at the lesson-level) decision.