Smart Videos Coming Soon
Videos move us. They fill our screens with sight, sound, motion, color, culture, and humanity. Videos educate, engage, connect, persuade, and entertain. But if you think about it, today’s web videos are “dumb”. They don’t reveal themselves to machines or to humans without consuming them! Videos are not easily searched and indexed. Which means it’s difficult to get only the relevant videos of quality when you search. To understand whether a video is what you want, you have to watch it, so videos can waste our time. And they are difficult to consume in any context, any language, or in segments.
First steps to “smart” video are delivered by captions
Enter Captions, which make videos smarter in multiple ways:
- When included in page, caption transcripts are indexed by search engines, lifting the videos search rank for relevant and specific terms, and for research.
- Captions enlarge the audience to ESL and hearing-impaired viewers (about 20% of the population) and to anyone viewing without audio, for example in a public or mobile context.
- Captions have been shown to increase audience engagement, time spent, and video completion.
- Captions enable navigation, chaptering, and search within via an interactive transcript, where the viewers can navigate by text.
- Captions will soon be legally required on much of the world’s web videos.
- Captions pave the way for low-cost language translation and more language enabling.
Translations take captions a step further by making videos available and more discoverable to new audiences. 77% of YouTube’s video (for example) are viewed non-native English language viewers. Companies like dotSUB enable both time-coded captioning and translation in any of the world’s language (currently 514 languages have been completed). Translation can be performed by any combination of professionals, employees, curated crowd-sourcing, volunteers, or machines.3rd step to “smart” video will be to derive time-coded keywords, meta-data, and abstracts from video transcripts
- Contextual search – Video search will become as good or better than text and site search. This will save time, aid research, and increase learning and satisfaction.
- Curated playlists – Web Video is evolving to longer sessions. Curated programming will prevail as it now does in the audio realm.
- Hyper links – Web video will inherit all the semantic associative playfulness of the web.
- Advertising – Only relevant ads will be viewed, at more reasonable intervals, in more entertaining and engaging formats – a win for both brands and viewers.
- Abstracts or snippets – Will allow audiences to browse to decide what to watch, and will enable new media mashups.
Google recently won a patent on a technique that identifies objects in a video, living or not. Google proposes using a database of “feature vectors” such as color, movement, shape and texture to automatically identify subjects in the frame through their common traits. Video makers can jumpstart the underlying material just by naming and tagging some of their clips, with these more accurate labels helping to separate the wheat from the chaff for the automated visual ranking system. This method would enable Google to fill in YouTube keywords without any user intervention — a potential boon to advertising relevance and video playlist curation.