Back in 2004, I would watch as summer interns (people who wants access, any access) spend sleepless nights as they watched a bank of televisions and VCRs and transcribed the important notes that would be useful in the morning notes that was sent to the Senior staff. This task, not a welcome one, should have been easily handled by some computer or automated procedure. But it took quite some time until it became more real.
Enter 2006, as I was doing work for GoodnightBurbank, I learned of a technology called tveyes that would automatically transcribe content and provide the transcript in a fashion that would make it more easily searchable by google. There were some issues I had with it regarding the comedy podcast (worried that the keyword richness would not be as helpful), but always remembered this for the next campaign I would help.
Then, recently, I got an email from a company called EveryZing asking to talk to me about their technology. As the discussion began, I commented on the technology's similarity to tveyes. It turns out that EveryZing IS the technology that powered tveyes - and they are heading out on their own. This got me to look again, and decided to investigate what you might do with it.
DISCLOSURE: this was a blogger relations call their firm set up and, while I have not personally used the technology yet, I can see the applications for campaigns as well as other efforts.
Their opening email
In the initial email, they sent me following:
As we've witnessed with the Revered Wright controversy and Obama's speech on race that followed - it is crucial that voters have access to valuable speeches and other political information on the Web. It's a part of how they make decisions, form opinions...and vote.
It's well known that online searchers typically only go to the first page of results to find what they're looking for - but how does a political campaign or media company make sure that their video is found, that they're meeting voter demand, and that they're receiving the Web traffic and ad dollars they deserve?
EveryZing leverages its pedigreed speech-to-text technology, based on breakthrough Department of Defense R&D at BBN Technologies, to unlock the content within multimedia. EveryZing is the only service to create a full-text output of search results of audio and video, allowing voters to easily find the specific content they want no matter where it lies within the file. Not only do you find that video of Hillary Clinton's Kentucky speech when you Google it, but you can also use EveryZing's "jump-to" technology to go exactly to the point where she says "they deserve to have those votes counted."
Their three products actually cover the tools that I would want for a campaign - specifically if I was trying to extend the value of my video assets into a search environment. For example:
Example: New video to highlight a topic
Once a video is produced, the video file can be uploaded to the EveryZing ezSEO application (I assume installed on your site) which will then provide the metatagging necessary for the video content and then submit the content to the video and text search engines. Why does this matter? Because in the world of the google PageRank algorithm and "freshness" metric, video is becoming more powerful than simply text, since it requires a lot more on the production side and clicks/views are a hearty metric on association to the PageRank algorithm.
Why does this matter? Simple - when video is produced, the expense needs to be spread across as many mediums as possible - audio can be reused for radio ads, and I can tell you the countless times people tell me to "put the video up on the web" and then have it languish there since no one could find it, unless they were directed toward it. But with metatagging and transcription, the video can be found any number of ways - and with EveryZing's inline tagging, you can find the when and why people are saying what they have said with incredible accuracy.
Future: sentiment analysis and rapid response
One of the discussions I had with Tom (the CEO) was the issue behind sentiment analysis and how you could determine if the videos that were transcribing had a positive or negative sentiment. Since machine learning has progressed in such a way, it should be an easy tool to develop. Once that is in place, I could see campaigns installing their own automated research applications (even down to the local level) and tracking TV and radio for sentiments. Once that is set with an alert feature, rapid-response can be incredible.
I look forward to seeing what happens next - and best of luck to EveryZing.