Video streaming stress caused by slow buffering could be banished to the dark ages, thanks to new AI technology developed by researchers at MIT. Could Netflix and YouTube be about to change for the better?
Most video streaming works using something called an adaptive bitrate algorithm, which loads video in discrete chunks and selects an appropriate bitrate – the rate of data transmission, or download speed – for the job. Slow internet, for instance, is dealt with by lowering quality for patchy periods. If internet coverage drops for too long, viewing is interrupted completely.
But MIT hopes to improve on this system even further, thanks to a new custom-built artificial intelligence. Pensieve, as the AI is known, uses network information combined with information about the state of the current buffer to predict periods of slower internet. Consider commuting through tunnels on a train or drops in internet speed in internet cafes.
The aim is to avoid the dreaded pixelated soliloquies of Francis Underwood in House of Cards or outright outages of The Mexanines latest tracks on YouTube, by preparing your content in a more efficient way.
Rather than block buffering a chunk of video or audio at a time, the system will pragmatically load more or less of the media as required.
The system was trained with combinations of buffer loads and network speeds. Users judged the quality to be higher and video playback experienced less rebuffering.
There’s no telling whether popular video apps will adopt such a technology any time soon, but the fact that Pensieve even exists gives us hope for the future of media streaming.
Let us know what you think of this story via Facebook or Twitter.