Identify time offsets for diachronic assets

[cproctor]

For each of the following:

• Zoom audio recordings
• Other audio recordings
• Minecraft Replay videos
• Audio transcripts
• Backend analytics

We need to specify the process by which a user could select a time interval from the file. (This is more of a writing-and-thinking task than a coding task) Following this, we need to identify what kinds of metadata will be needed. For example, we definitely need metadata for correcting the server's logged timestamps with an offset. We also need metadata for each audio recording specifying the UTC datetime when it starts.

[remcmanu]

Zoom audio recordings / transcripts

• at least looking back at past things I've zoom recorded, there's a timestamp to when the recording was taken, which you'd just add to the timestamp in the transcript. So if Chris says "hello" at 1:30 (a minute and a half into the recording), and the recording when he says that shows the time as 6:31:30pm, you know he started the recording at 6:30 on the dot.
• Should be really easy to make into UTC, IF the zoom recording is doing what it did for me back during the semester.

Other audio recordings:

• This one was by far the hardest. I essentially looked for moments I could find logs for (players movements, chats, etc), and tried to sync it down best I could. I figure I could get within a few seconds with this method, but it'd be more helpful to have the metadata on Chris' device where he recorded this. Since I pulled from Github, I think the timestamp is from then.
• Should be really easy to make into UTC, IF we can get the time from the file. If not, it'll be moderately difficult.

Minecraft Replay videos

• I look in the log files for the recorder's Join message (as it starts recording as soon as you enter the game)
• For example:
  • Viewed in Minecraft Replay, the file shows it started recording at 5:58 on 7-7-21. According to Chris' log [[2021-07-07T21:58:41.0067][PlayerJoinEvent]: "player": "thispettypace"] he joined at 21:58:41.0067, aka 5:58:41pm.
• Should be really easy to make into UTC.

Backend analytics

• this is just UTC. so 22:00 is 6:00pm

[cproctor]

This is good progress. Let's start a specification:

• For each raw data file, source.ext, we'll optionally have a metadata file source.ext.meta.yaml in the same directory.
• For now, we'll define the following optional keys:
  • begin: A timestamp for the beginning of the file. This is required for data formats which have a notion of time duration built in. For example, audio and video. For data formats which specify absolute timestamps (logs, some transcripts), begin is optional and overrides the starting timestamp, so that the effect is to shift all the timestamps. begin should be in YAML timestamp format, e.g. begin: !!timestamp 2021-07-14 13:51:02.4.
  • offset: A time delta by which to offset absolute timestamps in data files. Data files which encode absolute timestamps may use begin or offset, but not both.