Subtitles with an ampersand fail to work correctly

[dagwieers]

Describe the bug

I noticed this before, but failed to report it. If subtitles contain an ampersand, the remainder of the line gets cut off. Example: Finale Campus Cup (episode 24) around 5:50 CD&V is mentioned, only CD is being printed.

To Reproduce

Steps to reproduce the behavior:

1. Go to Campus Cup Finale (episode 24)
2. Forward to 5:50
3. Watch CD&V being cut off

Expected behavior

It should show subtitles correctly.

Additional context

• Operating system: LibreELEC 9.0.2
• Kodi version: 18.2
• Addon version: 2.0.0
• Using a VPN: no
• Country you are using the addon from: BE

[mediaminister]

This is not an inputstream.adaptive bug. It's again a VRT NU bug.

The same bug occurs on the VRT NU-website: https://www.vrt.be/vrtnu/a-z/de-campus-cup/1/campus-cup-s1a24/

I extracted the TTML from the stream:

ffmpeg -i "https://remix-vrt.cdn.eurovisioncdn.net/remix/f0bf9f98-5d65-43b4-84df-d85d05a7444e/remix.ism/.mpd" -map 0:7 -c:d copy -copy_unknown -f data campuscupfinale.ttml

TTML Fragment:

<p begin="00:05:38.453" end="00:05:40.693" region="region-11" tts:textAlign="center">
<span style="singleHeightStyle" xml:space="preserve" tts:backgroundColor="black">Vingers aan de knoppen. Vraag 1.</span>
</p><p begin="00:05:40.773" end="00:05:46.333" region="region-10" tts:textAlign="center">
<span style="singleHeightStyle" xml:space="preserve" tts:backgroundColor="black">Belgische partijen. We kennen de </span><br></br>
<span style="singleHeightStyle" xml:space="preserve" tts:backgroundColor="black">grote partijen zoals CD</span>
</p><p begin="00:05:46.413" end="00:05:48.293" region="region-11" tts:textAlign="center">
<span style="singleHeightStyle" xml:space="preserve" tts:backgroundColor="black">Enfin, vroeger grote partijen.</span>
</p>

The same bug occurs with HLS and WebVTT, disabling inputstream.adaptive can't fix this: https://remix-vrt.cdn.eurovisioncdn.net/remix/f0bf9f98-5d65-43b4-84df-d85d05a7444e/remix.ism/remix-textstream_dut=1000.vtt

WebVTT Fragment:

00:05:38.453 --> 00:05:40.693
Vingers aan de knoppen. Vraag 1.

00:05:40.773 --> 00:05:46.333
Belgische partijen. We kennen de 
grote partijen zoals CD

00:05:46.413 --> 00:05:48.293
Enfin, vroeger grote partijen.

We can't fix a bad subtitle source. This bug should be reported with "Meld een probleem" on https://www.vrt.be/vrtnu/help/

[dagwieers]

~I will report it to VRT.~ Reported!

[mediaminister]

I just found out that the same bug occurs with Terzake at 6:01: https://www.vrt.be/vrtnu/a-z/terzake/2019/terzake-d20190704/ It's definitely a bug in VRT NU's subtitling back end.

[dagwieers]

VRT acknowledged the problem and forwarded it to the responsible team.

[mediaminister]

Not fixed yet: https://www.vrt.be/vrtnu/a-z/terzake/2019/terzake-d20190923/#autoplay=1054&asset=/content/dam/vrt/2019/09/23/terzake-20190923-ma-depot_WP00148139

[dagwieers]

Correct, and I asked for an update last week from VRT (with CD&V being in the news and all subtitles being cut off) but VRT did not have any new updates from the engineering team.

[mediaminister]

7 months later and not fixed yet: timecode 24:57 in https://www.vrt.be/vrtnu/a-z/terzake/2020/terzake-d20200129

[dagwieers]

I reported it 3 times already, and it was confirmed 3 times. Feel free to report it once more :wink:

[dagwieers]

Still an issue, example timecode 29:55 in https://www.vrt.be/vrtnu/a-z/terzake/2020/terzake-d20200504/

[dagwieers]

@nielslaukens I reported this to VRT NU a few times the past year. Are you aware of this subtitle issue?

[nielslaukens]

We were aware of a few issues regarding subtitles, but this one wasn't explicitly mentioned. It could have been part of another issue, though. I've reached out to the corresponding team and gave this a bump, although that is not guaranteed to have any effect.

[nielslaukens]

This got escalated to our encoder vendor. They accepted this as a bug, but have not provided a timeline for fixing this. They did provide an (elaborate) workaround, which we may implement in the mean time. Keep you posted.

[mediaminister]

Just a reminder, this is not fixed yet: Terzake 2020-10-27 timecode 00:02:06.130 https://www.vrt.be/vrtnu/a-z/terzake/2020/terzake-d20201027/

[michaelarnauts]

Misschien eens doorsturen naar [email protected] dat haar partij geviseerd word door de VRT.

[dagwieers]

I reported this once more to the VRT Helpdesk. It would be nice if we can get this fixed before the 2-year anniversary of this ticket 😉