Audio descriptors for Transport Stream

[Marcelo Naumann Boufleur]

Current ffmpeg multiplexing using "-f mpegts" does not create audio descriptors for AC3 audio.

Original stdout for some transport stream content with descriptors are shown like this:

Stream #0:1[0x1e2](por): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 448 kb/s

After remultiplexing the same content through ffmpeg using this command

$ffmpeg -i INPUT.ts -map 0:0 -map 0:1 -map 0:2 -c:v copy -c:a copy -f mpegts OUTPUT.ts

ffmpeg outputs the audio in transport stream container, and is then again read by ffmpeg as this:

Stream #0:1[0x101](por): Audio: ac3 ([129][0][0][0] / 0x0081), 48000Hz, stereo, fltp, 448 kb/s

The original descriptors for audio are lost during re-multiplex. The original descriptors are identified by "(AC-3 / 0x332D4341)" information.

When using the option "-mpegts_flags +system_b", ffmpeg outputs the TS file with the following audio characteristics:

Stream #0:1[0x101](por): Audio: ac3 ([6][0][0][0] / 0x0006), 48000Hz, stereo, fltp, 448 kb/s

This last option does seem to change stream type from 0x81 to 0x6 in order to comply with DVB standards and also creates an AC3 Descriptor (with value 0x6A) for the audio.

There are however cases when it would still be interesting to maintain ATSC standards, and apply audio descriptors in the final content.

The original content above has a Registration Descriptor (0x05) with the value of 0x41432D33 that is the format identifier of AC-3, and an ATSC AC-3 Descriptor (0x81) - which is within the (0x81) stream type.

There is a more complete information available on ffmpeg-user list, in the following link, with more detailed description of the original and resulting remultiplexed files using a transport stream analyser software.
​http://ffmpeg.org/pipermail/ffmpeg-user/2016-September/033536.html

[Carl Eugen Hoyos]

Sorry for the delay, this is a duplicate of ticket #2162 (that does have a bad title).