We have live HLS streams with separate aac audio for multi-language tracks.
Previously we have had 6 second segments, which had led to a small variation in audio segment length, although this played back everywhere on iOS14/tvOS14.
In iOS15/tvOS15 these streams fail to play, unless we align audio and video segment lengths, which in turn requires us to run at 4 or 8 seconds for segments.
To get back to 6, we would need to sample at a much higher rate (192kHz), as opposed to the 96kHz we have working today, or the 48kHz we had working previously.
Anyone else spotted this?