FFMpeg is the swiss army knife of video editors and motion artists worldwide. Let’s take a look at one of the lesser-known (and kinda unexpected) features.
FFmpeg can easily extract embedded subtitles from videos. This command will grab the default subtitle track and export it as a srt file:
ffmpeg -i input_file out.srt
What if we want to get a different subtitle? First we need to figure out the track number for that subtitle by running ffmpeg -i input_file. You will get an output that something like this:
[...] Stream #0:2(eng): Subtitle: subrip (default) Metadata: title : English-SRT Stream #0:3(eng): Subtitle: hdmv_pgs_subtitle Metadata: title : English-PGS Stream #0:3(chi): Subtitle: hdmv_pgs_subtitle Metadata: title : Chinese-PGS [...]
Notice the tracks are numbered #0:2, #0:3, etc. This is the value we want to pass over to the map command to select the proper subtitle.
ffmpeg -i input_file out.srt -map 0:3
Removing HTML from subtitles
There are a number of applications and online services that can strip html tags from subtitles (like HTML Stripper), but you can also solve this quickly with good old sed:
sed -e 's/<[^>]*>//g' subs.srt
You can make an alias to this in your favorite shell so you don’t have to remember or copy/paste it all the time. Keep in mind that the results of sed won’t be as good as a HTML stripper and company, since this is just a simple regex. That said, it’s usually more than enough for subtitles.
Converting to other formats
While FFmpeg can get this done for you with a limited number of formats (depending on how it was compiled) there is a better alternative. SubtitleEdit is a handy open source application that can convert between 200+ subtitle formats (and do a lot more of course).
If you want to bring those subtitles into Blender check out the SubsImport addon.