I listed the cases where yt-dl applies the fixup above. The uncertainty arises from yt-dl/ffmpeg not knowing how the output will be used, not from the actual "malformation" (a poor description of the situation, probably chosen by a speaker whose first language wasn't English).
Basically (see Wikipedia) streamed AAC (as pulled from HLS fragments) comes with self-synchronising ADTS headers, which an implementation decoding AAC from an MP4 container need not understand. The ffmpeg bitstream filter
My guess is that the Humax player s/w, which definitely has to be able to play M2TS files with ADTS, also accepts ADTS in MP4.
Basically (see Wikipedia) streamed AAC (as pulled from HLS fragments) comes with self-synchronising ADTS headers, which an implementation decoding AAC from an MP4 container need not understand. The ffmpeg bitstream filter
aac_adtstoasc in the fixup transforms the streamed ADTS-AAC audio into the MPEG-4 Audio Specific Configuration for AAC, which an implementation decoding AAC from an MP4 container is required to understand.My guess is that the Humax player s/w, which definitely has to be able to play M2TS files with ADTS, also accepts ADTS in MP4.
