Blinkx to Contextualize Video Ads Through Speech Recognition
June 22, 2007 by Mike Abundo
Filed under Computers
Currently, video ad contextualization is achieved through text metadata — i.e., descriptions and tags. Unfortunately, text metadata is highly susceptible to keyword spamming, as anyone who’s ever run a popular search on YouTube can confirm. Video search engine Blinkx plans to use something that’s harder to spam, unless you’re willing to say random irrelevant words in your video: speech recognition.
Blinkx is putting the final touches on a new advertising platform for Web video code-named Project Trilby. When it launches next Monday it will be called blinkx AdHoc. It will offer media companies and video sites a way to place targeted ads alongside (or even in) Web videos based on the specific words spoken in the videos, as well as their overall context.
Web video advertising is the fastest-growing segment online, yet nobody really knows how to make those ads as relevant as search ads. Blinkx CEO Suranga Chandratillake says:
You need something akin to AdSense for video, which is what we have built.
Blinkx already indexes more than 12 million hours of video on the Web, and makes it searchable via speech-to-text and clustering technologies. Now, with AdHoc/Project Trilby any large Web video producer will be able to send their videos to blinkx to be indexed so that when somebody watches that video a targeted ad can be triggered.
Looks like Kevin Nalts’ joke video plugging startup, Cash to Buzz, might have a future after all.
Google just announced their own video fingerprinting technology. Wonder if they’ll extend that into video recognition to compete. After all, they practically invented contextual advertising through AdSense.
Nevertheless, I’d like to see where Blinkx can take their technology. In any face-to-face group conversation, certain audio keywords always catch peoples’ attention: “free”, “money”, “sex”. It’ll be very interesting to see how videographers optimize their audio keywords for advertisers.
Comments
6 Responses to “Blinkx to Contextualize Video Ads Through Speech Recognition”Trackbacks
Check out what others are saying about this post...[...] Blinkx have a jump start on the speech to text technology but no indication that they see value in providing transcripts to viewers. The real focus for both Blinkx and Everyzing is providing targeted ads. Blinkx will launch Adhoc on Monday: “Just like Google’s AdSense matches search ads with the words on any given publisher’s Web page, blinkx will bring a similar form of contextual mapping to video. An advertiser can buy keywords, and the ads will be triggered when those words are spoken in a video or they appear in a title, description or a tag attached to the video. The ads can be delivered right back on the publisher’s site selected from their own inventory, that of an ad network, or from blinkx. “ via Inside Online Video [...]
[...] bye, Blinx AdHoc. We hardly knew [...]
[...] online video ad revenues are growing 55.5% per year. With various video-sharing sites racing to contextualize video ads and expand ad partnerships, expect that growth to only accelerate in the next few years. If [...]
[...] Blinkx launches video advertising platform. It uses “Blinkx’s patented speech-to-text transcription and visual analysis technology to understand video content” and automatically pick relevant ads. More information about the speech recognition technology. [...]
[...] #3: I can’t find relevant videos online! Nontext video search technology, based on audio and video recognition, is being built as you read [...]
[...] any popular video-sharing site. Appropriately dubbed AdHoc, the program contextualizes ads using Blinkx’s speech recognition technology. By contrast, every other program out there that puts ads in embeds relies on text keywords for [...]