search

clarin-eric / parla-clarin

Schema for modelling parliamentary debates
https://clarin-eric.github.io/parla-clarin/
21 stars 6 forks source link

Recording synchronization section #20

Open matyaskopp opened 2 years ago

matyaskopp commented 2 years ago

https://clarin-eric.github.io/parla-clarin/#sec-speechvideo I have a few points for this section

time in recording

https://github.com/clarin-eric/parla-clarin/blob/4db5213e0c3d9faf4276a3fa65503d78cb6e53d4/Schema/parla-clarin-odd.xml#L1760 I guess this should be the date of recording. I prefer element <date> with attribute @when to clarify it. or use @from and @to when DateTime value is used

linking audio and timeline

https://github.com/clarin-eric/parla-clarin/blob/4db5213e0c3d9faf4276a3fa65503d78cb6e53d4/Schema/parla-clarin-odd.xml#L1784 I think linking timeline with recording based on time is not very efficient (there is no other relation in data). Furthermore, there can be more recording and more timelines in data. In ParCzech 3.0, I used timeline/@corresp attribute that targets recording/@xml:id.

absolute time

https://github.com/clarin-eric/parla-clarin/blob/4db5213e0c3d9faf4276a3fa65503d78cb6e53d4/Schema/parla-clarin-odd.xml#L1784-L1785 This example is lifted out of context. So it can be misleading. Value @absolute should contain absolute time (midnight in this case). But it is not really absolute... I prefer using the DateTime format to be more concrete. Example from ParCzech 3.0:

<timeline unit="ms" 
          origin="#ps2017-091-01-005-005.audio1.origin" 
          corresp="#ps2017-091-01-005-005.audio1" 
          cert="0.889">
   <when xml:id="ps2017-091-01-005-005.audio1.origin" 
         absolute="2021-03-23T18:18:00"/>

timeline placement recommendation

timelines elements do not have recommended placement (ParCzech 3.0 placed it at the end of //body - after //body/div[@type="debateSection"])

Examples in this section

I think that examples in this section should describe the same period. (I do not expect that someone says Good morning, <anchor synch="#T2"/>Mr. president! one second after midnight.) It would help the reader to understand the connection between recording, timeline and u.