Web video has exploded over the past few years. The perfect storm of YouTube (and other video sharing sites), cheap video camcorders, rapidly growing broadband adoption (pdf), and a seemingly endless number star-wars kids and dogs on skateboards has created a web where video is becoming ubiquitous. Even long-term internet staples, like The Onion and MLB.com have added video content. And while some ISPs are arguing (pdf) that their tubes can’t handle the bandwidth from online video (an argument that is, incidentally, specious (pdf)), there’s no way to put the genie back in the bottle with regard to online video.
That having been said, for all that online video can do – illustrate something better seen than read, empower users to create their own content, or destroy politicians with their own words – it has its weaknesses. In particular, video’s not great for skimming or sampling, it’s tough to reference, and can be tricky to pass around at the office or classroom (well, unless you’re our office – these three videos got quite a bit of play at Beaconfire last week). In addition, it can be tough to search a video library for a particular section – particularly if it’s an audio reference, searching may have to be done in real time.
MIT has found a way to mitigate this last problem, and they’re using their own video lectures as a guinea pig. From MIT’s technology review:
Announced last month, the MIT Lecture Browser website gives the general public detailed access to more than 200 lectures publicly available though the university’s OpenCourseWare initiative. The search engine leverages decades’ worth of speech-recognition research at MIT and other institutions to convert audio into text and make it searchable.
This is simply an amazing innovation. More on why this could mean an explosion for online video in the future below the fold.
While at my last gig, I can recall the pain of transcription, which we did only on those pieces we really needed text of for searchability or transmission purposes. It could take hours to accurately create a transcript from a single speech. With MIT’s indexing technology, this labor could be a thing of the past.
Some would argue that there are few merits to turning a multimedia piece into one that’s text only. I’d respond by pointing out just a few potential benefits:
- Search engine indexing – As was discussed in the article, making content searchable makes it infinitely more useful. Look at the prevalence of Lexis-Nexus as a tool for finding a quote from a particular television program (I’ve discussed the power of Lexis previously as one of the only sites that can work as a paid service). This opens this complex and expensive solution to the masses, not just on the output end but the input as well. In addition to searching lectures and talking-head shows, you could use it to quickly find the episode of House with the final diagnosis of lupus. And the college game of "Name that Movie Quote" will never be the same.
- Text analysis – This is currently only performed on a narrow band of video – speeches that include prepared version or other critical texts that are examined by the masses in other ways already. But what if you wanted to how many times Jon Stewart does a Dick Cheney impression in a given month of The Daily Show? Or to find the phrases that are used most often in "Ask a Ninja?"
- Accessibility – We talk quite a bit about accessibility here, but video often moves sites in the opposite direction. Since there are both audio and visual elements, it’s more likely that you’ll leave people unable to fully experience your content. But almost anyone can handle text, either through a screen reader or magnification. Think of it as a poor NPO’s closed captioning.
- Mobile readers – whether it’s a Blackberry, Treo or even a new Kindle (incidentally, Amazon, thanks for advertising something that’s out of stock until post-Christmas on your homepage), your mobile device probably can’t do video. Even the iPhone’s vaunted "just like the real web" experience is only true if you just want your video from YouTube. Mobile devices can deal with text much easier.
Indexing of video could continue to advance Google’s domination of the world – right now, video remains one of the most difficult media to search, and with their acquisition of YouTube, Google could be positioned to take advantage of this technology. That said, YouTube is not really the ideal target for a deployment – much of their content is more about the visual than the audible, particularly given that the primary focus of YouTube is short and grainy stuff, as opposed to longer pieces that might see a greater benefit.
That said, however, they might be making the move away from this reputation – they recently released a multi-file uploader and increased the file size limit from 100mb to 1gb, meaning we could see longer and higher quality video there.
Video’s not going anywhere, and making it searchable and auto-transcribed will probably actually make it more prevalent, since content creators would not need to weigh its weaknesses as heavily anymore. When that happens, I promise you’ll see some more video blogging here. Until then, looks like you’ll have to be content with what’s on YouTube for now.