If every attendee is able to set which language they will be speaking during the meeting, then the captions and translation AI will automatically know what the input language is from that audio source and it eliminates the step for meeting hosts to change the spoken language in between speakers. This enables a seamless multilingual conversation while maintaining accessibility, and would reduce the likelihood of errors if (for example) a German speaker comments on a presentation that's being made in Chinese. (It also reduces the need for all meetings to be conducted in one language that is not attendees primary language).