Subtitle sync

[GerardCorbella]

Dear all,

we're working with Dédalo v6 in an oral history archive. For this project we have created a kind of exhibtion that shows very concret exceprts of interviews in 3 different languages.

We're using subtitles in the excerpts, but some times there is a small time delay between audiovisual play and subtitles starting.

Is it possible to refine?

[renderpci]

Hi GerardCorbella.

Of course, subtitles calculation can be improved. They are an automatic calculation of the fragments and in some situations the calculation is complicate and the result is not perfect, is an approximation.

Can you explain the problem with an example? what is the situation in the indexing and the result of the calculation.

Are you test if you insert a timecode close to the indexing tag? the result is the same?

Best

[GerardCorbella]

Dear all,

after fine-tuning putting timecodes close to the indexation tags, results are really best, and in the majority of cases it works properly.

Anyway, in some cases result can be better. For example, in this case, with timecode next to indexation tag, we have a 9 seconds delay with previous text in the subtitles: https://preservemlamemoria.coib.cat/dedalo/publication/server_api/v1/subtitles/?section_id=64&lang=lg-cat&tc_in=3652.692&tc_out=3672.625

[renderpci]

Hi GerardCorbella.

I can't see in the link the error that you mention.

The VTT begins at 00:00:00.000 with text and you have other 3 lines with 00:00:00.000 timecode and continue after that in the 9 second... but, it's not an error, in VTT you can define different text lines with the same timecode, in the parsed subtitles you will see this text lines at this time, all together...

Can you explain with more detail?
What is in the transcription at this point?

The time of your call is:

• tc_in = 3652.692
• tc_out = 3672.625

That is equivalent to:

• tc_in = 01:52.692
• tc_out= 02:12.625

and what should be the expected result in VTT?

Best

[GerardCorbella]

Hi,

here you have the excerpt published online: https://preservemlamemoria.coib.cat/cuidar/ca/respectant/5

If you compare with the VTT provided before, subtitles starts online in the fourth paragraph, but audiovisuals starts in the third one (with the interviewer). Delay it's minimal (around 2 seconds).

Basically what we expect is to have a perfect fit between audio and subtitles, we know that's difficult and 2 seconds delay it's practically perfect, and we're only sharing this issue with the aim to improve Dédalo in the next future.

Best,