Project Entangle is Koherence’s internal platform for exploring various media technologies. It began as a test bed for investigating how TiVo-style trick modes could be implemented using HTTP Live Streaming. OTA recording, program guide support, and content search came along shortly after. Deployed Entangle systems are currently being used as DVRs, with users watching shows on iPhones, iPads and AppleTVs.
Entangle is an ever-evolving platform. Some enhancements support investigations by Koherence’s clients. For example, to better assist a streaming service in crafting their long-term transcoding plan (and to do so with a bit of experience and credibility), Entangle’s HLS pipeline was refitted to support fragmented MP4 and H.265.
Others changes are in response to DVR users asking, “Why can’t I do [insert your feature here]?” For example, one request was to tackle the multi-antenna problem. One of the issues with OTA ATSC is that if you put an antenna at one spot in the attic and orient it properly, you’ll get one set of stations. Put another antenna six feet away, and you’ll get a slightly different set. Even more exacerbating, the set of channels can move depending on the weather or time of day. And to top it all off, there may be a time of day when reception slowly changes from one antenna to the other. Of course your favorite show will be airing during that transition time. You’d need to change antenna inputs at some point, and mid-show neither signal would look particularly good. Wouldn’t it be nice if you were somehow able to combine the signal from those two antennas? Then you could get all the channels all the time. And maybe there’d even be a way to tackle your favorite show that’s transitioning between antennas by taking the bits received well by each antenna and stitching them back together .
As we move beyond typical DVR and streaming features, there’s a host of technologies on the Entangle drawing board. ATSC 3.0 is right on the horizon and Entangle will be ready to handle all those 4K H.265 broadcasts. And we’re just starting to imagine the possibilities that deep neural-network processing of media can provide: speaker identification, text-to-speech, actor identification, scene description. It may not be too far in the future where Entangle can derive incredibly rich metadata by “watching” a show, and maybe even telling you which new shows you just have to see!
If you’re interested in running an Entangle platform, want to contribute to its continued evolution, or have a feature request, get in touch with Mike.