Disclosed herein is a computer-implemented method, system, device, and computer program product for securing private audio in a virtual conference. For each of the users in the virtual conference, a device of the speaking user or a server determines whether a respective user is able to hear the speaking user based on whether a respective sound volume at which the respective user is able to hear the speaking user exceeds a threshold amount. The speaking user's device or the server prevents the transmission of the audio stream to devices of the users determined not to be able to hear the speaking user.