DENVER — This week’s StreamTV Show provided an opportunity to learn about a couple of new technologies in the video streaming space.
First, an engineering track touched on improving latency of video streams through cutting-edge software techniques.
Tomas Bacik, head of client solutions at the company CDN77, talked about a couple of ways that the company’s video processing and delivery software is reducing latency for streaming video.
The CDN77 network is built with a mixture of strategically located PoPs and cache servers deployed directly inside ISP networks, all interconnected with a global private backbone as well as multiple Tier 1 transit providers.
At the StreamTV Show, Bacik first talked about HTTP chunk fetching as a way to speed video processing.
He said the company’s software breaks video segments into smaller “chunks” and sends those HTTP video chunks for processing even before they’re fully created.
“We need to teach the encoder to serve as many small chunks as possible and pass them along,” said Bacik. “The same theory applies to storage. You need to teach your storage to serve these incomplete files.”
He explained that breaking HTTP video into the shorter chunks increases efficiency, but “the real strength comes when you combine this with pre-fetch.”
He said, “For livestreaming we initiate the pre-fetch even before the segment is created. When the cache receives those chunks from segment 1 it can immediately request from segment 2 even before segment 2 exists. It keeps the request open and waits until the chunk is created. This allows us to get ahead of end users.”
CoCoPIE.ai
On the sidelines of the show, the CEO of a startup called CoCoPIE.ai was talking about the company’s artificial intelligence software that improves the resolution of streaming video via an app.
CEO Jay Liu said CoCoPIE.ai is based in Boston and has about 10 employees. Its app can take low-resolution video that’s coming from a CDN and upscale it to high-resolution video at the device level. The secret sauce is CoCoPIE.ai’s AI software.
The company sells its product to streaming platforms, which then load the AI software onto their applications through regular software upgrades.
Liu said CoCoPIE already has one major customer, China-based iQIYI, which he referred to as “the Netflix of Asia.”
But now, he’s hoping to sell his technology to American streaming platforms — and that’s why he was talking to everyone he could at the StreamTV Show.
The value proposition is that streaming platforms could subscribe for low-resolution video through their CDN provider and then boost the resolution of the video for the end users via CoCoPIE.ai’s technology, saving money on networking costs. Also, the AI software could improve resolution in situations with varying bandwidth.
Meanwhile, a couple of operators on another StreamTV Show panel indicated that subscribers now routinely expect 4K video resolution, but in reality, most people couldn’t distinguish a difference between 4K and HD on their TVs.