A new system and method particularly useful in handling the audio of a “large” conference is disclosed. The system and method initially defines certain conferees as panel members, which can provide audio to and receive audio from the conference, and audience members, which only passively receive audio from the conference. In one aspect, an audience member can request to participate in the conference. By making an appropriate request, the audience member can, if system resources are adequate, participate fully as a panel member and may supply audio to the conference for at least some portion of time. Alternatively, the audience member can supply information other than audio data to the conference, such as voting data or other data allowing the audience member to participate in the conference as more than a merely passive listening participant.