WebVTT -> SRT conversion #25

New Issue

Barbagus · 2023-01-05T19:46:18Z

Barbagus commented

2023-01-05 19:46:18 +00:00

Exploring #24, I found more information about FFMPEG's WebVTT support and why we could not manage to mux the subtitles without converting them to SRT.

It turns out FFMPEG as a very poor support for WebVTT and stop processing the data as soon as it find's something it doesn't understand, the most problematic for us being styling wich is always used by ArteTV.

This issue has been submitted to FFMPEG and a patch was submitted but it did not get traction since October 2010. Which confort me to think converting to SRT is the way to go for us.

HOWEVER loosing style information is a pitty, especially for textual description (subtitles for people suffering from hearing impairement) that uses colors to identify between descriptions and transcriptions. But it looks like some SRT styling is kind of supported by most players through the use of <font color=""> tag.

To actually handle that sort of convertion (WebVTT styles to SRT font tags), it seems though that the webvtt-py would not help a lot (their parsing is rather primitive). This added to #24 that comes from the fact that webvtt-py is too strict for us... well... looks like implementing it from scratch could be the way to go.

Exploring #24, I found more information about FFMPEG's WebVTT support and why we could not manage to mux the subtitles without converting them to SRT. It turns out FFMPEG as a very poor support for WebVTT and stop processing the data as soon as it find's something it doesn't understand, the most problematic for us being _styling_ wich is always used by ArteTV. [This issue has been submitted to FFMPEG](https://trac.ffmpeg.org/ticket/9064) and a [patch](https://patchwork.ffmpeg.org/project/ffmpeg/patch/CA+_k4RJ1CNriPbhxxzLf1O6V1VuP2OXch3E8LqOsMNSRUyEnTw@mail.gmail.com/) was submitted but it did not get traction since October 2010. Which confort me to think converting to SRT is the way to go for us. _HOWEVER_ loosing style information is a pitty, especially for textual description (subtitles for people suffering from hearing impairement) that uses colors to identify between descriptions and transcriptions. But it looks like some SRT styling is kind of supported by most players through the use of `<font color="">` tag. To actually handle that sort of convertion (WebVTT styles to SRT font tags), it seems though that the `webvtt-py` would not help a lot (their parsing is rather primitive). This added to #24 that comes from the fact that `webvtt-py` is too strict for us... well... looks like implementing it _from scratch_ could be the way to go.

Barbagus referenced this issue from a commit

2023-01-06 00:22:47 +00:00

Fix #24 and #25

Barbagus referenced this issue

2023-01-06 00:24:35 +00:00

Fix #24 and #25 #26

Barbagus closed this issue

2023-01-06 00:26:07 +00:00

Sign in to join this conversation.