Is it possible to extract SubRip (SRT) subtitles from an MP4 video with ffmpeg?
I have checked the FFMpeg documentation and many forums and figured out the correct command-line to extract subtitles from an .MP4 video should look like so:
ffmpeg -i video.mp4 -vn -an -codec:s:0 srt out.srt
However, I get the following error, which lends me to question whether this is feasible at all:
Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height
Using ffmpeg -codecs, I can confirm that ffmpeg should be able to encode subrip subtitles.
Using ffmpeg -i video.mp4, I can see that there is two subtitle tracks embedded in the video :
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'video.mp4':
...
Stream #0:0(und): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p, 720x572 [SAR 64:45 DAR 256:143], 1341 kb/s, 25 fps, 25 tbr, 90k tbn, 180k tbc
Stream #0:1(eng): Audio: aac (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 191 kb/s
Stream #0:2(fra): Audio: aac (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 191 kb/s
Stream #0:3(eng): Subtitle: dvd_subtitle (mp4s / 0x7334706D)
Stream #0:4(und): Subtitle: mov_text (text / 0x74786574)
EDIT
I have tested with the simplified command-line shown in the comments but I still get the same error. Here is a link to the detailed verbose output from running the command. I have also tried to completely disable metadata and chapters in the resulting output but that still produces the same error.
Answer
I enventually figured out why I did not succeed:
The specified command-line would have been perfectly fine if the subtitles from the source video were encoded in a text-based representation. However, as can be seen in the output to the ffmpeg -i command-line, the subtitles are encoded in the "dvd_subtitle" format.
The dvd_subtitle format stores bitmaps for each subtitle in the video. Therefore, there is no way ffmpeg would be able to translate the bitmaps into text.
For this task, one has to resort to an OCR-based software which assists a user with the task of identifying each subtitle as text from its bitmap représentation.
(There is a secondary text-based subtitle in the source video, but I don't know where it came from and is not seen by most popular players. For all intents and purposes, this "mov_text" subtitle seems to be a stub placeholder, probably an artifact of the conversion from the original DVD)