DTX stands for Discontinuous Transmission.
DTX is an extension to the Opus voice codec and can be enabled via SDP negotiation.
With DTX enabled, if there’s silence or just background noise that is being encoded by Opus, it will be replaced with a single frame every 400 milliseconds instead of the default of a frame for every 20 milliseconds.
DTX is a useful tool for large group calls where the SFU needs to route many audio channels to the viewers. Since usually one or two people will be speaking at a time, any other participants whose microphone isn’t muted can be switched to DTX mode, saving on bandwidth and packets.