We released the first beta version of JackTrip 2.2 earlier this week for testing purposes. Among other things, this release includes latency optimizations when using JackTrip’s default buffer strategy (loss concealment).
This work was inspired by previous test results that indicated the “minimal latency” strategy offered slightly lower latency. Today, we re-ran tests from Palo Alto to Portland (about 560 miles) using the newest releases of JackTrip. Our earlier tests used JackTrip 2.0, and here are the latest results:
Product | One-way Latency | Details |
---|---|---|
JackTrip 2.1.0 | 36 ms | 128 Buffer Size, Loss Concealment |
JackTrip 2.1.0 | 30 ms | 64 Buffer Size, Loss Concealment |
JackTrip 2.2.0-beta1 | 33 ms | 128 Buffer Size, Loss Concealment |
JackTrip 2.2.0-beta1 | 25 ms | 64 Buffer Size, Loss Concealment |
Note that these results represent one-way “mouth to ear” measurements, including all the network transit, digital-analog conversions and buffering that happen along the way. Most applications (including JackTrip) only report network transit times in-app.
As always, your actual latency is largely going to depend on your home Internet connection (just use Fiber, if you can), and how far away you are located from one another. Everyone will get different results. But in general, most people using 2.2 who follow all of our recommendations (most importantly, plugging into Ethernet!) should expect a 3-6 millisecond reduction in latency out of the box.
Other updates in 2.2 include improvements across the board when you are using a different input device (i.e. microphone) and output device (i.e. headphones). These also apply to everyone using audio built into their laptop. Even though we don’t recommend it, it should work better. It’s also likely that latency will be significantly lower when using different audio devices (especially on Windows), even by as much as 10 milliseconds or more.