Re: [Audacity-devel] Add support for importing/exporting SRT subtitles as labels

[Pokechu22 <pok...@gm...>]

I've created pull request #394
<https://github.com/audacity/audacity/pull/394>, which is a rework of my
old PR (#383) <https://github.com/audacity/audacity/pull/383> combined with
some ideas from #170 <https://github.com/audacity/audacity/pull/170>.

--Poke

On Wed, Nov 13, 2019 at 7:11 PM Robert Hänggi <aar...@gm...>
wrote:

> On 14/10/2019, Pokechu22 <pok...@gm...> wrote:
> > For me at least, I implemented this so that I could caption an audio file
> > <https://commons.wikimedia.org/wiki/File:Eisenhower_farewell_address.ogg
> >.
> > Captioning a video file would definitely also work, but that adds a bit
> of
> > complexity regarding positions on screen (which both SubRip and WebVTT
> > support to various degrees).  I don't think it's super important for the
> > average user (since I don't think many people use labels as is) but I
> think
> > people who have a use for it would benefit greatly.
> >
> > Barring that, one thing I did not address with SubRip is that it supports
> > multiple lines in captions, which is not supported by labels in audacity.
> > Beyond just editing, the current text format would not work with multiple
> > lines on a single label.  With the current PR I just flattened it into
> one
> > line when importing, but more advanced labels would allow for it.
> >
> > Code-wise, I would like some advice on how to improve the timestamp
> > parsing, since it's rather messy.
> >
> > Regarding Karaoke: it probably failed to record since it's not rendered
> > using the GPU.  An external screen recorder (e.g. OBS
> > <https://obsproject.com/>) probably would work.  It probably also would
> be
> > possible to render a video using FFmpeg since I believe you can pipe raw
> > video to it and it can encode it, but I don't know enough about that to
> > implement it.
> >
> > Regarding images: I don't think any subtitle format supports those, since
> > it would not be machine readable.  It looks like WebVVT might support
> > it through
> > CSS <https://w3c.github.io/webvtt/#obtaining-css-boxes>, but it would be
> > pretty messy.  I'm not entirely sure how it'd be used, either.
> The images would constitute the video stream.  It is fairly simple to
> use FFMPEG to combine images and audio. However, it would be even
> simpler by using a label track to place those images exactly.
> Ffmpeg allows also to integrate a subtitle stream, thus you could use
> a second label track to feed in this information.
> Alternatively, the labels could be rendered in order to produce lyrics
> videos. It would be advantageous if Audacity could render the Karaoke
> window content directly as you wouldn't have the frame rate issues due
> to slow computers.
> My view point is that we have now more or less track abstraction that
> would allow to introduce types that are derived from the label track
> type.
> It is also thinkable that you use the track dropdown menu of a label
> track to check the radio button "Subtitles" or "Lyrics" or "Playlist"
> which would inform Audacity what to do on export.
> Anyway, just a little fancy of mine.
> It is a good start Poke if you just implement the SRT export option.
> Robert
>
> > --Poke
> >
> > On Mon, Oct 14, 2019 at 12:00 PM Robert Hänggi <aar...@gm...>
> > wrote:
> >
> >> I'm 100 % for it.
> >> While we're at video:
> >> I've tried to make a lyrics video from Audacity by screen capturing
> >> the Karaoke  window (with the XBox game bar, Windows+G) but it didn't
> >> work for my graphics card.
> >> A native way to do it would be great.
> >> It would also be nice to be able to import still images into a label
> >> and let Audacity export that.
> >> Looking forward to your suggestions/PRs.
> >> Robert
> >>
> >> On 14/10/2019, James Crook <cr...@in...> wrote:
> >> > We're not doing too well on staying on top of pull requests.
> >> > We have 26 open and 338 closed.
> >> >
> >> > Ideally we'd have one person responsible for ensuring every pull
> >> > request
> >> > got an answer and most of the time have no open pull requests.
> >> >
> >> > I suspect nothing happened there because no one here had worked with
> >> > sub
> >> > rip format, so noone had an idea of how important (or not) it was.
> >> > I don't think support for WebVTT is essential to progressing/accepting
> >> > srt support.  I do think if we intend WebVTT, it would be wise to plan
> >> > an update to labels.
> >> >
> >> > --James.
> >> >
> >> >
> >> > On 14/10/2019 18:49, Jonatã Bolzan wrote:
> >> >> Hi
> >> >>
> >> >> I see no discussion about this PR:
> >> >> https://github.com/audacity/audacity/pull/170
> >> >>
> >> >> What do you think? I think this is interesting as Audacity can be
> used
> >> >> for
> >> >> video-related projects. In the future, I would be interested in
> >> extending
> >> >> support for WebVTT if this is something you are interested in.
> >> >>
> >> >> Thanks for considering.
> >> >> --
> >> >> Jonatã Bolzan Loss
> >> >>
> >> >>
> >> >>
> >> >> _______________________________________________
> >> >> audacity-devel mailing list
> >> >> aud...@li...
> >> >> https://lists.sourceforge.net/lists/listinfo/audacity-devel
> >> >
> >> >
> >> >
> >>
> >>
> >> _______________________________________________
> >> audacity-devel mailing list
> >> aud...@li...
> >> https://lists.sourceforge.net/lists/listinfo/audacity-devel
> >>
> >
>
>
> _______________________________________________
> audacity-devel mailing list
> aud...@li...
> https://lists.sourceforge.net/lists/listinfo/audacity-devel
>