I am constructing VOD HLS streams out of S3 with captions. The captions are loaded from a TTML file and converted to WebVTT format in the HLS stream. We are seeing a problem when the TTML contains captions with overlapping begin/end times, but with different positions. For example:
<p begin="01:01:39.491" xml:id="p8" end="01:01:43.704" region="pop1" xml:space="preserve"tts:origin="15.00% 79.33%"style="basic">
<span tts:fontStyle="italic">Line One</span>
</p>
<p begin="01:01:39.491" xml:id="p9" end="01:01:43.704" region="pop2" xml:space="preserve"tts:origin="12.50% 84.67%"style="basic">
<span tts:fontStyle="italic">Line Two</span>
</p>
In the above, I would expect to see “Line One” placed above “Line Two” on the screen.
When Wowza converts this to VTT, it looks like this:
00:01:35.491 --> 00:01:39.704
<i>Line One</i>
00:01:35.491 --> 00:01:39.704
<i>Line Two</i>
The result is that players show only “Line Two”. In order for this to be rendered properly, I would expect the output to look more like this:
00:01:35.491 --> 00:01:39.704
<i>Line One</i>
<i>Line Two</i>
Does Wowza support conversion of TTML captions this way? Is there anything we can do to work around this problem?