Example embodiments of the present disclosure provide techniques for multiple audio device capture and playback during a remote desktop session. A custom device manager may instantiate proxy audio input and output devices on the server corresponding to each real device on the client computer. The proxy devices may redirect their respective audio input and output data to the physical device on the client. In an embodiment, multiple audio devices may be instantiated during a remote session, and multiple independent audio streams may be redirected to the selected device on the client computer.