Create a script to postprocess videos

[ligi]

currently it is quite a manual process and the people that can do it are quite busy. But I think we can automate it with a script which consists of these 3 steps:

1. replace camera audio with audio from microphones
2. cut out sections and add metadata
3. add intro / outro

I think 3 is trivial with ffmpeg. Replacing the audio track also - just sync would be the tricky thing here - also wondering if a clap would still be needed or if there are fancier ways already - the audio tracks from both sources with random input that is recorded anyway should be enough info to sync already. And IMHO the trickiest part will be to find the cutpoints for 2 - as far as I see there will be the need for a human in the loop - but we can reduce the amount of work for the human by reducing the possible start times to short sections by using a trigger from e.g. the camera operator to find the rough starting time.

Optional (after the MVP):

• customize the intro outro as was done for older meetups (e.g. contain month/year) - but I think for the beginning static intros / outros are also OK - for the start IMHO a static intro/outro are good enough - we could take the intro here - just need to find a way to remove the date/year here.

[chrishobcroft]

Another approach to this is to evolve the pre-production / setup to make post-production as minimal as possible.

For example:

1. can be done in pre-production by connecting the mic audio into the camera, so that the camera records the mic audio into the video file. This is not without it's complexities however, as some cameras cannot take an audio input, plus it requires either audio cables, or some kind of RF transmitter (e.g. Rode Link).

2. can be done with a signal from the event host, to whoever is managing the camera to "start recording now", and "stop recording now", at the start / end of an individual talk.

With these changes, it can be possible to end the event with 1 video file per talk, to be fed into step 3, which as stated in the initial comment, is just a case of some ffmpeg scripts.

[ligi]

Thanks for chiming in! Unfortunately connecting the sound directly to the camera is not that easy - running cables in this setup is problematic - and we do not have a second RF link (also not sure about the reliability of 2 RF links) so I think the "recording at the source" is the better option for now - also think merging them should be easy and this way we also have data redundancy.
Regarding (2) this sounds good - we need to try it out - can just imagine to find the correct start/end times on the spot can be hard.