I'm receiving an Opus audio stream, and occasionally I receive packets which consist of 49 zeros, followed by x02. Reading the Opus RFC, I see that the table-of-contents (TOC) byte being zero should correspond to SILK-only transmission with a frame size of 10 ms, but I don't think that's what these packets are.
Another weird thing is that these Opus packets have RTP headers, and the RTP timestamp for these zero packets seems to be equal to the timestamp for the preceding timestep. Looking at the RFC for the Opus + RTP format, I don't see anything obvious which would explain this.
These packets don't seem to be corruption in the data stream, the Opus decoder handles them just fine. Anyone know what might be going on?