A media transmission method based on telepresence is disclosed, comprising: during connection establishment, a master telepresence terminal of a local media transmission system and a remote endpoint perform signaling exchange, to establish a media logical channel between the local media transmission system and the remote endpoint; and transmits the same type of media streams between the local media transmission system and the remote endpoint through a media logical channel or respectively through multiple media logical channels, and receives the same type of media streams through a media logical channel or respectively through multiple media logical channels. A media transmission system based on telepresence is further disclosed correspondingly. Through the disclosure, when a telepresence system of a certain conference site is called, only one number needs to be called, which is easy to operate; moreover, agent information exchange can be performed in the telepresence system, thereby achieving the effect of sound-based localization, solving the problem such as synchronization between streams, and improving user experience.