This incident has been resolved, and latencies should be back to normal for the last 2 hours now.
It seems that the root cause is one of our provider having issues with bandwith limitations, resulting in some large audio struggle to get passed between our services. We're still investigating to make sure it doesn't happen and we're putting mitigation in place for this.
If some of your request have failed due to timeout issues, those have been automatically replayed.
