Your environment.

• Version: latest

What happened?

The bug lies in this section of code in the Kalman filter..
https://github.com/pion/interceptor/blob/master/pkg/gcc/kalman.go#L81-L84
I think it should be an if else statement, as this is what I'm seeing in another implementation https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/blob/main/net/rtp/src/gcc/imp/kalman_estimator.rs?ref_type=heads#L53-59
and how I interpret it from the RFC

If z(i) > 3sqrt(var_v_hat) the filter is updated with 3sqrt(var_v_hat) rather than z(i).

In https://datatracker.ietf.org/doc/html/draft-ietf-rmcat-gcc-02#section-5.3

It would be great to have another pair of eyes on it