Anthony Rose, chief technology officer of social TV app Zeebox, on the recent addition of automatic content recognition, the biggest technical challenges involved, and what he sees as the defining features of a successful second-screen app.
Anthony will be speaking at the OTTtv World Summit taking place in London on 19-22 November 2013. For more information and to register for the event, please visit ottworldsummit.com.
Zeebox has taken the major step of adding automatic content recognition. Can you talk us through your aims here?
In Zeebox our goal is really to do two things – help you find programmes to watch, and also to help you have a great time while you’re watching. So to help you have a great time while you’re watching, we started off designing Zeebox as a companion to live television, and I think some people read that as ‘Anthony moving from iPlayer where it as all about on-demand to Zeebox where it is all about live, does this mean that video-on-demand is no longer the future?’ I think people read far too much into it.
We started Zeebox around live television for three reasons: firstly of course, millions of people watch live TV; secondly, they do it at the same time, so it creates a social community that is a bit fragmented when people are watching video-on-demand, specifically around social; and thirdly, it’s technically easier, because you don’t have to deal with the problem of figuring out where in the show an individual user is when they are using a PVR.
But, times move on, and more and more people are watching video-on-demand, and our US partners tell us that for drama and comedy and scripted content, more than 50% of viewing is on-demand – and this is clearly a trend that is rising.
That doesn’t mean that all genres are moving to on-demand – obviously news, sports and reality shows continue to be primarily viewed live – however, more and more drama is viewed on-demand. That was problem number one.
Two, in the US and Australia where we are now live, there were multiple time zones in between the east and west coasts of both countries, creating the problem of Twitter spoilers, where people don’t like to watch the Twitter stream for recorded or time-shifted shows, because they get to see what happened at the end of the show.
So how could we solve this? For west coast viewing of live TV, the answer is that the schedule says you are watching a show three hours later, so you don’t need automatic content recognition. But for PVR viewing, you need to recognise how far you are into a show, and for that we use automatic content recognition.
It is important to know that we use ACR to synchronise the experience in Zeebox with the play point in the show that you are watching, meaning that we have the ability to replay the entire show experience, including votes, polls, games and even the Twitter stream, synchronised to the viewing experience – and I think that is unique and Zeebox is the first to do that.
That is fantastic in bridging the gap between the live, social, everyone-is-doing-it-together experience with the lonesome PVR viewing experience: you may find that in future you will be able to watch a show when you want, and still have a social experience around it.
In due course, when we link to set-top boxes and remote recorded content, you could invite your friends to have a viewing party and all watch together with a replayed social experience, brought in synch with the show just for you.
Now I should point out that for at least the last year, Zeebox has used video fingerprinting done on our servers to recognise ads on TV – we use server-side video ad recognition so we can synchronise commercial advertising and other experiences to the show that’s on TV. What we are doing now is something different and additional – which is to use client-side audio content recognition to synchronise your viewing experience with the experience that was already created for that show.
Zeebox has also added a new recommendations engine – can you tell us a little about why this was done?
In terms of recommendations, as I mentioned earlier we at Zeebox split our focus on helping you find things to watch, and giving you a great time while you watch it. What we want to do now is turn Zeebox into your personalised programme guide.
We all know that there is more and more content available, more and more channels available, and a traditional channel guide on a cable or satellite set-top box is crazy – you can get repetitive strain injury trying to find something to watch.
A better scenario is to follow some shows or channels, and when you pick up your tablet or smartphone it says ‘hey Anthony, Top Gear starts in ten minutes, Glee is on right now’, and when you tap on it then Zeebox becomes a remote control and tells your set-top box to play that show.
What’s fascinating is the question of ‘where does this progress’? In future, if your tablet or phone has a list of shows for you, and you click on it and it tells your set-top box or TV to play that show, you won’t care where it came from – over an aerial, or from the PVR, or over the Internet. Essentially we will see a complete blurring of content sources.
Now some people say that recommendations are just based on things that the viewer already knows about, and use that as a critique of recommendations as a source of viewing, but actually I don’t think that is a valid criticism. In fact, a good recommendation engine should tell you about things that you already know about, and things that you don’t yet know but hopefully will like.
I look at it as three concentric circles – the inner circle covers shows that you know and love, the outer circle contains shows that you don’t know and are not going to like, and the interesting middle circle (which a good recommendation engine should address) contains the shows that you don’t yet know about but would like.
Pages: 1 2