NBC looks inside the video to put the Olympics online
Posted on August 18, 2008
Chris Albrecht from NewTeeVee recently posted an article about some of the ways the Olympics have pushed the envelope of online video technology. NBC has made over 2,200 hours of video available on NBCOlympics.com.
As Chris points out, Microsoft and Digital Rapids were forced to innovate themselves out of some unique challenges related to how much hardware space was available to NBC in the Olympic broadcast center:
“To help fit in those constraints, Digital Rapids and Microsoft came up with a new technology called Dynamic Complexity that scales the amount of encoding power from the CPU based on what is being shown. For example, with an actual sporting event, you want to make sure all of those action-packed bits get transmitted, so the video doesn’t get compressed as much. For something more sedate, like video of a talking head, you don’t need as many bits, so you can compress the video more. The Dynamic Complexity helps even out the loads being carried by the bandwidth.”
The basic idea here is that what the human brain considers “very high quality” is not strictly dependent on bitrate alone; the actual content of the video combined with the bitrate determines what is considered great quality. Watching Bob Costas, as a relatively stationary figure, can be delivered at a lower bitrate while still “looking” like very high quality video to the viewer.
This is another good example of how additional intelligence about the video itself, by looking inside the video, can lead to better results and why publishers need to look past legacy video platform providers for next generation solutions. I expect to see more innovation like this from ourselves and others in the future.
Alex
Delve Partners with EdgeCast to deliver high quality video
Posted on July 21, 2008
Today’s we’ve announced that EdgeCast has been selected to provide CDN services for the Delve video management platform. After carefully evaluating many alternatives, there were a few reasons why we decided to work with EdgeCast:
- Domain Knowledge - our engineering team can be quite demanding and has an incredibly high bar for itself and the vendors they rely on. Edgecast distinguished itself by demonstrating a great deal of technical competency.
- Reliability - reliability is a key issue for our customers, which means it is a key issue for us, EdgeCast leverages backbone infrastructure from several of the largest telecom providers all over the world.
- Performance - we were impressed with EdgeCasts ability to provide high quality video content through their vast network of caching location. This, along with our own infrastructure, will help ensure our publishers can deliver the highest quality video watching experience to their audience.
We like the EdgeCast team and we’re happy to work with them; however, our philosophy has been to ensure reliability and performance. As a result of this desire, we will have relationships with several CDNs as well as maintaining own infrastructure across multiple data-centers.
Alex Castro
Lies, Damn Lies, and Video Metrics
Posted on July 19, 2008
This week, ComScore came out with online metrics for May. Here are some of the highlights:
- 12 billion videos were viewed during the month of May in the US
- This represents an increase of 45% compared to a year ago
- Google sites (mostly YouTube) represents 34.58% of all videos watched
- There is a huge drop-off from there, with Fox Interactive (MySpace Video) coming in second with 6.4%
- Hulu breaks into the top 10 with 0.7% share (88 million videos)
Unfortunately, ComScore leaves a number of questions unanswered:
- How does ComScore define “video site”? Does this include sites that focus exclusively on video like YouTube, Hulu, MySpace Video, etc? Or does this include video that augments a site like newspapers, magazines, and other Internet sites (e.g. content, community, etc.)?
- What is a “video server network” and how is it different that a “video content site”? This is an important questions since ComScore explains that their metrics are for “video content sites” only and exclusive “video server networks.”
- If we understood how they are defining these terms, we could determine how much video consumption is from video oriented sites, and how much video consumption is coming from other web sites that increasingly include video. Basically, is video becoming ubiquitous across web sites across the Internet?
- How does the top 10 list change if ranked by total time spent watching video? I would be surprised if YouTube still wasn’t #1 by this measure, but would Hulu (with longer form content) still be #10?
- Where are sites like Veoh, MetaCafe, Heavy, and DailyMotion? These are all top 100 sites according to Alexa; much higher than Hulu; yet the don’t show up anywhere on this list. Is this because ComScore focused on viewers in the US?
My take away is that video metrics, as with many things related to online video today (e.g. advertising) there is a lot of confusion and ambiguity. While reports like these from ComScore are certainly helpful, they do not yet provide the comprehensive data necessary to bring clarity to this market.
In order for publishers, advertisers, etc., to make intelligent decisions about video they will need to look at available metrics from sources like ComScore and others, but they will also need to make some projections based on previous historical trends for how previous Internet trends evolved and matured.
Alex Castro
Goodbye Brightcove, Hello Brightcove
Posted on June 18, 2008
Today, Brightcove has acknowledged their product lacks the ease-of-use and advanced capabilities publishers require to be successful. Or, as TechCrunch so aptly puts it, “Brightcove Gutted and Rebuilt.” After reading the various articles and blog posts, I came to the conclusion that there wasn’t much substance to this release and even less forward innovation available for publishers.
Here is a summary of their new functionality:
- Making the user interface less tedious and complicated.
We’ve heard from many Brightcove’s customers that the user interface
was productivity killer, and Brightcove’s poor user experience is a key
reason they are evaluating Delve.
- Making it a easier to move out of the “video ghetto.”
With Brightcove, a publisher’s high quality video is forced to live in
the “video ghetto.” This is relegating video to one small corner of a
website. Delve’s ability to extract meaning from video gives publishers
more powerful tools to contextually integrate video content throughout
their website.
- Improving poor video quality and delivery. This
is another concern we’ve heard from Brightcove’s customers. At Delve,
we are fortunate enough to have a team of developers with deep
experience building “Internet Scale” services at places like
Amazon.com, Amazon Web Services, Yahoo Search, Inktomi, Expedia, MSN,
etc. This is a core part of our DNA, which is different than that of
Brightcove’s where most of the senior technical leadership came from a
legacy software background.
Publishers must ask - “Which online video platform will help me succeed?”
In the online video market innovation occurs at a blistering pace. Publishers need a partner capable of keeping up with, and most importantly staying ahead of, the industry trends. By the time Brightcove’s new release reaches the market, they will have spent 18+ months on it. However, if you look at their website, they haven’t delivered much additional value to existing customers since since April 2006 - an incredibly long 30 months in a fast moving market like video.
While the market demands innovative features like search, video SEO, enhanced video management, higher quality video, enhanced advertising, increased user engagement, Brightcove has been playing catch-up with basic features like providing a user interface that meets the basic usability needs of users and copying innovations introduced by their competition. This puts their customers at a disadvantage in the marketplace.
I was also very surprised to see Brightcove attempt a complete rewrite - they are basically throwing out their entire publishing system and starting from scratch. If there is one thing I’ve learned is that there is a tremendous amount of risk associated with complete rewrites. One doesn’t need to look much farther than Microsoft’s challenges with Windows Vista for evidence of this.
Ironically, the complete rewrite approach results in the evaporation of Brightcove’s main selling point to customers - “we’re market tested.” Since Brightcove’s new release can no longer claim to be “market tested”, publishers have to really evaluate video platform providers on the basis of who is innovating, who is bringing them additional value more quickly and on a regular basis, and who will ultimately help them succeed in this space.
I firmly believe that Delve Networks provides the most advanced, intelligent, and comprehensive video platform. We don’t just give publishers basic features like transcoding and a video library, but instead provide a whole host of more advanced capabilities (e.g. search inside) derived from our core technology and ability to look within the video content itself.
Alex Castro
CEO
Delve Networks
The Delve Video Platform
Posted on June 14, 2008
By Alex Castro
Delve Networks
Web video has never been more popular, as evidenced by the fact that over 11.2 billion online videos were viewed in the United States in March of 2008, according to ComScore Video Metrix. Yet, you have to wonder if the current video platforms have kept up with demand—both on the publisher side and user side of the web video viewing equation.
Consumption has grown at a geometric rate over the past few years and it’s imperative that the technology for distributing and viewing that video—along with monetization through advertising—keeps pace with user demand for rich media content.
So, what are the top ten things that content companies should look for in a video platform?
1. A reliable hosting solution that scales with their business.
2. A fully featured video player— with normal play controls, plus sharing and posting functions.
3. An advertising solution that allows the publisher to monetize their video assets, either by placing their own ads or integrating with third party ad services (or both).
4. Metrics— enabling the publisher to rapidly measure the consumption of their video and learn more about their web-based audience.
5. An easy way to upload and manage video files to the video platform. This should be a secure, web-based interface that does not require advanced technical skill.
6. A consumer-facing solution that preserves and enhances the content publisher’s brand. This can be done by “skinning” the player to match the publisher’s web site and corporate “look and feel.”
7. The ability to easily organize videos into channels or playlists.
8. A video format that supports Hi Definition output.
9. A platform that is flexible enough to support company needs, but not so complicated that it requires special training or an inordinate investment of time to get going.
10. Great customer service. A publisher deserves to expect prompt access to the video platform team when any issues arise.
At Delve, we just made an exciting announcement about our new Video Platform. Obviously, we hope that we are raising the bar in our industry, helping publishers get more content into the ecosystem and giving users a radically new way to experience video content—giving them a chance to navigate within a file, before hitting “play.”
The innovations continue . . .
Neilsen Gets it Right: Minutes Matter Most on the Media Web
Posted on July 11, 2007
Nielsen’s announcement yesterday marks something a lot bigger than just a way to measure the relevance of sites that use AJAX. While folks like Steve Rubel correctly predicted the death of the page view metric some time ago , this new announcement does more: it lays down the foundation for measuring and monetizing the Media Web. This is a great example of the difference between the “fetch-a-page-of-text” web and ad model and the emerging interactive Media Web driven by live, time-variant experiences, like video.
Video context (and user intent) changes from moment to moment, creating additional monetization opportunities. The longer you keep the user around, the more opportunities you have to monetize, if your advertising approach is sophisticated enough to take advantage of that engagement.
One way to keep users around is to give them opportunities to interact with and control their media. Early experiments at DoubleClick shows that users interact with video at an astonishing rate, up to 29%. With our HearHere and SeeHere technology we give users the ultimate incentive to interact and engage, the ability to find what they want, and jump to it.
Once again, giving users control and convenience pays off: In our early deployments we’ve seen providing search increases user time with the media by 200%. For users that actually do search within the show, that goes up to 300%. This means potentially 2-3x more monetization.
Nielsen’s new measurement approach will become even more significant once advertising systems and models catch up. Virtually all ad business on the Internet is still focused on fetch-a-page-of-text, the entire business model was designed when page views still mattered. In this scenario, ad context is determined once and fixed for each show or clip when it is retrieved from the server. This is why, assuming a status quo approach to advertising, webpages seem to monetize better than videos, as this post on NewTeeVee points out.
At Pluggd, we change that by providing additional opportunities to monetize as video context changes and the user interacts and searches. And withSeeHere and HearHere, we help keep the user around to see those ads, because as this announcement from Nielsen points out, minutes are what matter most on the Media Web.
Hungry? Get your Hot CNet Chunks!
Posted on June 30, 2007
Yesterday, Pluggd and CNet unveiled our HearHere search capability over a sampling of popular ZDnet shows.
It’s exciting, because this is the first time people are able to search within media using a heatmap experience.
Before this, users had two choices: full meal deal or get up from the table.
Now they can find and snack on the tasty chunks they are looking for.
Try it out for yourself by doing a search for “iPhone” or “Apple” on this episode:
http://blogs.zdnet.com/BTL/?p=5411
Check out what some folks are saying about the HearHere release:
http://blog.seattlepi.nwsource.com/venture/archives/117373.asp
http://venturebeat.com/2007/06/29/pluggd-begins-delivery-of-better-audio-search/
http://mashable.com/2007/06/29/pluggd-launches-audio-search-player-on-cnet/
More to come!
Pluggd in the Economist
Posted on June 8, 2007
We’re excited to be mentioned in a story in the Economist today about
speech recognition. The article does a good job of surveying the space, but
what really makes Pluggd different (and speech reco useful for video search)
is the chunking technology we’ve developed (read more about chunks).
It’s great to see the company getting highlighted this way in such an
outstanding publication.
- Alex Castro
More about Chunks: The Parts You Want
Posted on May 30, 2007
We got some questions about what these chunk-things are after my last post. Some folks asked how this is different from just searching for the utterance of a word in video. It’s quite a bit different. Matt Marshall at Venture Beat did a good job describing how Pluggd works in this post (http://venturebeat.com/2006/12/06/pluggd-perfects-audio-and-video-search-raises-165m/) after we last spoke with him.
Let’s dig into this a little more by investigating user intention. When a user searches within video for the word ‘golf’, are they thinking, “The person who created this video has really good enunciation, I wonder how they pronounce the word ‘golf’?” I don’t think so. This is the type of user experience enabled by using speech recognition by itself.
Instead, the user’s intention is more likely to be, “I am really interested in golf, find me the segment within this video where golf is talked about.” This requires identifying a distinct and relevant conversation, what we call a ‘chunk’, within the video. Speech recognition alone isn’t enough to accomplish this. We combine speech recognition with some very interesting semantic analysis and information retrieval techniques to identify chunks. We are able to identify a chunk by recognizing when related words and word phrases (e.g. golf, Tiger Woods, green, Vijay Singh, under par, over par) are used in sequence within an area of video.
There are several interesting implications of chunking:
1) Far superior results than speech recognition by itself
Because we are using the presence of related words, as opposed to the presence of a single word, we are able to achieve results that are far superior to even the best speech recognition engines.
The diagram below illustrates how this works for a scenario where a user searches for a chunk by typing in the query term - “Vijay Singh.” The word phrase “Vijay Singh” might prove difficult for a speech recognition engine, including the one we use, to identify. However, our chunking technology compensates for this.
2) Increases a user’s media consumption
Because users can jump to exactly what they are interested within the video, they don’t ‘bail out’ of the video. Users often start watching a video clip only to become frustrated when they don’t immediately see what they were expecting, and they are too impatient to wait for the video segment they do care about. They just leave. Our experiments show that a very high number of users ‘bail out’ of video within the first 30 seconds.
By allowing users to jump to what they are interested in, users become satisfied, and spend more time watching more of the video. In fact, we’ve found evidence that users display some of the ‘browsing’ behavior in video that they exhibit with hyperlinks and text web pages. In a future post, I will share empirical data from some of the AB testing we’ve conducted over the past few months.
- Alex Castro
Google Combining Video with Search Results: Halfway There.
Posted on May 17, 2007
Google’s uber search across text, images and now video has a lot of people talking, including the New York Times. But the enthusiasm ignores how people really use search.
Searching includes not only finding results, but also figuring out which results you want. Google’s text search does a good job of highlighting a web page to help you determine which results fit your needs. People then “ping-pong” back and forth between search results and web pages to find what they’re looking for. This works because most of us can scan text very quickly.
Applying this experience to video is a disaster. You shouldn’t have to spend 3 minutes (the average length of a YouTube clip) watching a video to see if it is the one you want. In our research we’ve found the median viewing (or listening) time for a show is less than 10 seconds; when people don’t see what they want, they move on. So if you’re looking for a gem that is buried a minute into the show, forget it. It just doesn’t get found.
At Pluggd, we chunk audio and video. What’s a chunk? The part you want. We are the only ones that let you jump right to the tasty chunk you are looking for. A three minute video can feel like an eternity, which is why so many people bail in that first 10 seconds. We just don’t think you should have to sit around and wait for the good parts.
Pluggd is on a mission to chunk all video and audio for bite-sized consumption. More to come.
- Alex Castro
keep looking »