One of the things that attracts me to serialised feeds (as well as confusing the hell out of me) is the possibility of letting people subscribe to, and add, comments in “relative time”…
… that is, as well as viewing the content via a serialised feed, the comments feed should also be serialised (with timestamps for each comment calculated relative to the time at which the person commenting started receiving the serialised feed).
Applying this to the idea of tweeted Youtube movie subtitles (Twitter Powered Subtitles for Conference Audio/Videos on Youtube) in which every tweet made during a presentation at or around that presentation becomes a subtitle on a recording of that presentation, it strikes me that a similar model is possible.
That is, different individuals could watch a Youtube video at different times, tweeting along as they do so, and then these tweets could be aggregated according to relative timestamps to provide a single, combined set of subtitles.
So how might this work in practice? Here’s a thought experiment run through…
Firstly, it’d probably be convenient to set up a twitter account to send the tweets to (say @example, for example).
Create a tag for the video – this could be something like #yt:tBmFzF8szpo for the video at http://www.youtube.com/watch?v=tBmFzF8szpo.
(Alan Levine reminded me about flickr machine tags earlier today, which are maybe also worth considering in this respect, e.g. as a source of inspiration for a tagging convention?)
Grab a ctrl-C copy of the phrase @example #yt:tBmFzF8szpo for quick pasting into a new tweet, and then start watching the video, tweeting along as you do so…
To generate your subtitle feed, you can then do a search based on Tweets from your username (which would be @psychemedia in my case) to e.g. @example, with hashtag #yt:tBmFzF8szpo, and maybe also using a date range.
(You could augment the Yahoo pipe I used in the twitter subtitle generator proof of concept to remove the hashtag when generating the feed used for subtitling?)
The actual subtitle file generator could then pull in several different subtitle feeds from separate people, relativise their timestamps relative to the time of the first tweet (which could maybe use a keyword, too – such as “START”: @example START #yt:tBmFzF8szpo;-) and then produce an aggregated subtitle feed.
As more people watched the video (maybe including the subtitles to date), their feeds could be added to the aggregating subtitle file generator, and the subtitle file updated/refreshed.
Individuals could even rewatch the video and create new feeds for themselves to join in the emerging conversation…
(Okay, so it’s maybe slower than just reading through the comments, having to replay the video in real time to read the tweets, but this is a sort of thought experiment, right, albeit one that can be implemented quite easily…;-)
PS In one of the comments to Show and Translate YouTube Captions Matt Cutts gave an example of a URL that “will search for the word “china” in videos with closed captions” [ http://www.youtube.com/results?closed_captions=1&search_query=china ] (although I’m not sure how well it works?).
So I’m thinking – if live tweets from an event can be associated with a video of an event (maybe because the video is posted with a link to a (now out of date!) upcoming record for that event in order to anchor it in time) then being able to search the tweets as captions/subtitles provides a crib for deeplink searching into the video? (But then, I guess the Goog is looking at audio indexing anyway?)
PPS I just came across another tool for adding subtitles to Youtube videos, as well as videos from other online video sites – overstream.net:
It’s worth looking at, maybe?
PPPS see also Omnisio, a recent Google acquisition that offers “select clips from videos you find on YouTube and other video sites, and easily post them on your profile page or blog. Even better, you and your friends can add comments directly in the video!”.
And there’s more: “With Omnisio you make and share your own shows by assembling clips from different videos.” Roll on the remixes :-)