bobbed859 I personally don't recall it happening aswell on regular calls without recording, though 'not important' call usually don't reach the 10 min mark in my case.
Based on the tests i did above, based on call 4, it seems it might not be reproducable reliably. This might have a reason, but these are assumptions.
Call 3 & 4 are similar, though only call 3 invoked the crash.
On call 3 & 4 it was replicated only 1 partie was 'speaking', while the pixel was 'listening'. Though my kids where running around and being loud, so pixel might have had some audio input, but not actively.
It might be on call 3 there was enough audio input , and not on call 4. Based on call 5 & 6 which crashed several times when there was active audio input, this assumption might make sense.
It could be the issue lays there somewhere, combination of call recording, and active audio input in the pixel (perhaps combined with speaker).
Though these are very rough assumptions based on the limited data, and could even be unrelated at all. Nevertheless an interesing observation.