Techniques are described for offloading the encoding and decoding of multimedia content transmitted during a real-time communication (RTC) session from the virtual desktop to the client device so that the multimedia content can be communicated directly between the client device and the remote peer device without the involvement of the virtual desktop. The offload eliminates the additional network hop of the multimedia content to the virtual desktop that is present in conventional virtual desktop environments, thereby reducing network latency and improving performance, CPU utilization and network load on the virtual machine hosting the virtual desktop. In order to offload the multimedia encoding/decoding, the techniques described herein intercept a number of Application Programming Interface (API) calls to override the messages used to establish a multimedia communication session between the RTC application and the remote peer device.