The latency is dynamically set from the max latency of each audio output (reported by the OS or the WebAudio context) plus 200ms. I haven't tweaked this yet to

The latency is dynamically set from the max latency of each audio output (reported by the OS or the WebAudio context) plus 200ms. I haven't tweaked this yet to find the best possible value so it's still quite high. In the future I want to observe the network quality (packets loss mainly, latency on the local network is not an issue in most cases) to use a custom latency for each network. As I use Opus for the audio encoding, I've enabled the packet loss concealer which is already helping in case of package loss.