tympanix

Concerning problem 2 with "commercial breaks". I wrote, funnily also named [subsync](https://github.com/tympanix/subsync), a while back. I performed a recursive matching algorithm, splitting the audio/subtitle track into two halves each time,...

@smacke no worries about the name from my side of things. The approach worked very well for me with the tests I did. The matching was very reliable even with...

I agree. Finalizing the `Entry` should be transparent to the developer. But with no gab between the actual `Entry` and the `HandleLog` function, figuring out a solution is tricky. Could...

Sadly there is no docker image. I think it would also be troublesome because some users require older/newer versions of tensorflow to run the application. Also, since the application is...

Yes, maybe a small web UI would be cool 👍

We are invoking `ffprobe` to find the length of the video file here: https://github.com/tympanix/subsync/blob/661bc6856c6e428ec115bac85e053050d17a7fc1/subsync/ffmpeg.py#L55-L60 Apparently the output of the `ffprobe` command is something unexpected and therefore it does not match...

Maybe we can make the `ffprobe` invokation more robust with some tricks mentioned in the [documentation](http://trac.ffmpeg.org/wiki/FFprobeTips#Formatcontainerduration)

Ok, that is a problem. Do you get a non-zero exit code? Maybe the video file is unsupported by ffmpeg or it is corrupt in some way 🤔

@dependabot rebase

@dependabot rebase