Error in streamed MPEG4 handling - Movie.pm or Audio::Scan error ?

[philippe_44]

Clause 4.2 Size = 0 is a special case - it means all data up to eof

aligned(8) class Box (unsigned int(32) boxtype,
  optional unsigned int(8)[16] extended_type) {
  unsigned int(32) size;
  unsigned int(32) type = boxtype;
  if (size==1) {
      unsigned int(64) largesize;
  } else if (size==0) {
// box extends to end of file
   }
  if (boxtype==‘uuid’) { 
      unsigned int(8)[16] usertype = extended_type;
  }
}

But it's supposed to be 'size' first then 'boxtype', so I don't see what these 8 following zero are...

[bpa]

But it's supposed to be 'size' first then 'boxtype', so I don't see what these 8 following zero are...

Yes - I posted too quickly.

From your previous post

00 0D 6E F2 6D 64 61 74 00 00 00 00 00 00 00 00 00 D0 35 03
             m  d  a  t

Audio size is 880,370 which is 0DF26D - so size is OK, 'mdat' is OK.

Maybe sample table offset (or similar) has something interesting.

I'll look at it again tomorrow with fresh eyes.

[bpa]

Quick status on the problem.
I was trying find out where Audio::Scan gets its value for audio_offset (i.e how does Audio::Scan know to skip the extra zeroes in the mdat)
Looking at the MPEG4-Spec for mdat - the 2nd para in 8.1.1.1 for mdat

There may be any number of these boxes in the file (including zero, if all the media data is in other files).
The metadata refers to media data by its absolute offset within the file (see subclause 8.7.5, the Chunk
Offset Box); so Media Data Box headers and free space may easily be skipped, and files without any box
structure may also be referenced and used.

The following comes from the description for "stco" in clause 8.7.5

The chunk offset table gives the index of each chunk into the containing file. There are two variants,
permitting the use of 32‐bit or 64‐bit offsets. The latter is useful when managing very large
presentations. At most one of these variants will occur in any single instance of a sample table.
Offsets are file offsets, not the offset into any box within the file (e.g. Media Data Box). This permits
referring to media data in files without any box structure. It does also mean that care must be taken
when constructing a self‐contained ISO file with its metadata (Movie Box) at the front, as the size of the
Movie Box will affect the chunk offsets to the media data.

As I read it, the audio_offset from Audio::Scan is the value of the first entry in the chunk offset table.

I'm currently trying to do a quick fix in Movie.pm to test this out.

[bpa]

I have a patched Movie.pm which seems to play The problem URLs and a random m4a podcast which has a "normal" mdat. Need more example to test.

I used Audio::Scan seek_offset to update the "audio_offset" rather than first data in mdat.

/slimserver-public-8.3/Slim/Formats$ diff org-Movie.pm Movie.pm 
270a271,274
> 
>       $fh->seek(0, 0);
>       my $seek_offset = Audio::Scan->find_frame_fh_return_info( mp4 => $fh, 0 )->{'seek_offset'};
> 
277c281
<       $info->{audio_offset} = $args->{_mdat_} + 8;
---
>       $info->{audio_offset} = $seek_offset;

In parseStream routine

        # finally got it, add 'moov' size it if was last atom                                   # finally got it, add 'moov' size it if was last atom
        substr($args->{_scanbuf}, $args->{_offset} + ($args->{_atom} eq 'moov' ? $arg           substr($args->{_scanbuf}, $args->{_offset} + ($args->{_atom} eq 'moov' ? $arg

        # put at least 16 bytes after mdat or it confuses audio::scan (and header cre           # put at least 16 bytes after mdat or it confuses audio::scan (and header cre
        my $fh = File::Temp->new( DIR => Slim::Utils::Misc::getTempDir);                        my $fh = File::Temp->new( DIR => Slim::Utils::Misc::getTempDir);
        $fh->write($args->{_scanbuf} . pack('N', $args->{_audio_size}) . 'mdat' . ' '           $fh->write($args->{_scanbuf} . pack('N', $args->{_audio_size}) . 'mdat' . ' '
                                                                                      >
                                                                                      >         $fh->seek(0, 0);
                                                                                      >         my $seek_offset = Audio::Scan->find_frame_fh_return_info( mp4 => $fh, 0 )->{'seek_offset'}
                                                                                      >
        $fh->seek(0, 0);                                                                        $fh->seek(0, 0);

        my $info = Audio::Scan->scan_fh( mp4 => $fh )->{info};                                  my $info = Audio::Scan->scan_fh( mp4 => $fh )->{info};
        $info->{fh} = $fh;                                                                      $info->{fh} = $fh;

        # the offset is *after* the mdat atom, make size consistent                             # the offset is *after* the mdat atom, make size consistent
        $info->{audio_offset} = $args->{_mdat_} + 8;                                  |         $info->{audio_offset} = $seek_offset;
        $info->{audio_size} -= 8;                                                               $info->{audio_size} -= 8;

        # MPEG-4 audio = 64,  MPEG-4 ADTS main = 102, MPEG-4 ADTS Low Complexity = 10           # MPEG-4 audio = 64,  MPEG-4 ADTS main = 102, MPEG-4 ADTS Low Complexity = 10
        # MPEG-4 ADTS Scalable Sampling Rate = 104                                              # MPEG-4 ADTS Scalable Sampling Rate = 104
        if ($info->{tracks}->[0] && $info->{tracks}->[0]->{audio_type} == 64) {                 if ($info->{tracks}->[0] && $info->{tracks}->[0]->{audio_type} == 64) {
                $info->{processors} = { 'aac' => \&setADTSProcess };                                    $info->{processors} = { 'aac' => \&setADTSProcess };
        }                                                                                       }

        return $info;                                                                           return $info;

[philippe_44]

I have a patched Movie.pm which seems to play The problem URLs and a random m4a podcast which has a "normal" mdat. Need more example to test.

I used Audio::Scan seek_offset to update the "audio_offset" rather than first data in mdat.

/slimserver-public-8.3/Slim/Formats$ diff org-Movie.pm Movie.pm 
270a271,274
> 
>       $fh->seek(0, 0);
>       my $seek_offset = Audio::Scan->find_frame_fh_return_info( mp4 => $fh, 0 )->{'seek_offset'};
> 
277c281
<       $info->{audio_offset} = $args->{_mdat_} + 8;
---
>       $info->{audio_offset} = $seek_offset;

In parseStream routine

        # finally got it, add 'moov' size it if was last atom                                   # finally got it, add 'moov' size it if was last atom
        substr($args->{_scanbuf}, $args->{_offset} + ($args->{_atom} eq 'moov' ? $arg           substr($args->{_scanbuf}, $args->{_offset} + ($args->{_atom} eq 'moov' ? $arg

        # put at least 16 bytes after mdat or it confuses audio::scan (and header cre           # put at least 16 bytes after mdat or it confuses audio::scan (and header cre
        my $fh = File::Temp->new( DIR => Slim::Utils::Misc::getTempDir);                        my $fh = File::Temp->new( DIR => Slim::Utils::Misc::getTempDir);
        $fh->write($args->{_scanbuf} . pack('N', $args->{_audio_size}) . 'mdat' . ' '           $fh->write($args->{_scanbuf} . pack('N', $args->{_audio_size}) . 'mdat' . ' '
                                                                                      >
                                                                                      >         $fh->seek(0, 0);
                                                                                      >         my $seek_offset = Audio::Scan->find_frame_fh_return_info( mp4 => $fh, 0 )->{'seek_offset'}
                                                                                      >
        $fh->seek(0, 0);                                                                        $fh->seek(0, 0);

        my $info = Audio::Scan->scan_fh( mp4 => $fh )->{info};                                  my $info = Audio::Scan->scan_fh( mp4 => $fh )->{info};
        $info->{fh} = $fh;                                                                      $info->{fh} = $fh;

        # the offset is *after* the mdat atom, make size consistent                             # the offset is *after* the mdat atom, make size consistent
        $info->{audio_offset} = $args->{_mdat_} + 8;                                  |         $info->{audio_offset} = $seek_offset;
        $info->{audio_size} -= 8;                                                               $info->{audio_size} -= 8;

        # MPEG-4 audio = 64,  MPEG-4 ADTS main = 102, MPEG-4 ADTS Low Complexity = 10           # MPEG-4 audio = 64,  MPEG-4 ADTS main = 102, MPEG-4 ADTS Low Complexity = 10
        # MPEG-4 ADTS Scalable Sampling Rate = 104                                              # MPEG-4 ADTS Scalable Sampling Rate = 104
        if ($info->{tracks}->[0] && $info->{tracks}->[0]->{audio_type} == 64) {                 if ($info->{tracks}->[0] && $info->{tracks}->[0]->{audio_type} == 64) {
                $info->{processors} = { 'aac' => \&setADTSProcess };                                    $info->{processors} = { 'aac' => \&setADTSProcess };
        }                                                                                       }

        return $info;                                                                           return $info;

Thanks - I will also try to understand why my simple mdat offset calculation does not work

[bpa]

Thanks - I will also try to understand why my simple mdat offset calculation does not work

I think it assumed that audio data starts at first byte of mdat which if I read the stco explanation does not have to be the case.

Given that it seems to work when the Audio::Scan offset calculation is used, my explanation not be correct but the "fix" doesn't seem to break MP4 streaming. I have tested with a few MP4 audio podcast.

I don't use Tidal so will set up a trial account to test Tidal MP4 streaming to see if it still works.

[philippe_44]

I think it assumed that audio data starts at first byte of mdat which if I read the stco explanation does not have to be the case.

Given that it seems to work when the Audio::Scan offset calculation is used, my explanation not be correct but the "fix" doesn't seem to break MP4 streaming. I have tested with a few MP4 audio podcast.

I don't use Tidal so will set up a trial account to test Tidal MP4 streaming to see if it still works.

I will re-read the mpeg4 doc so see if I can easily put that inside the parseStream()

[philippe_44]

I've pushed that PR https://github.com/Logitech/slimserver/pull/749, let me know if this works for you

[mherger]

> I've pushed that PR https://github.com/Logitech/slimserver/pull/749

Thanks a lot philippe_44! I'll wait for bpa's thumbs up before merging. Ok?

[philippe_44]

> I've pushed that PR https://github.com/Logitech/slimserver/pull/749

Thanks a lot philippe_44! I'll wait for bpa's thumbs up before merging. Ok?

Absolutely!


Envoyé de mon iPad en utilisant Tapatalk