A teleconference between at least two participants is evaluated by monitoring the teleconference and detecting a triggering event if more than one participant desires to speak at the same time. Upon detecting a triggering event, e.g., detecting an attempt of one or more participants to interrupt another participant, or detecting a request initiated by one or more participants to interrupt another participant, a dominant source is selected from among the ones of the participants attempting/requesting to speak based at least in part upon a determined hierarchical order of those participants attempting/requesting to speak. A participant is then selected from among the participants to have the role of the dominant participant based upon the selected dominant source and only the speech of the dominant participant is conveyed to each participant through the teleconference system.