Advances in AI technologies like computer vision have changed the landscape of internet broadcasting. Before too long, algorithms will be better able to prevent sensitive and inappropriate content from being broadcast across the internet on livestreaming platforms

Nata Mollik
6 min readNov 7, 2020

The live streaming market is waiting to be tapped. Right now there are some prospectors, but in North America, no one has really found the vein leading to the mine. So maybe it’s time to get prospecting

Livestreaming, what’s in it for us?
Technology has advanced significantly since the first internet livestream but we still turn to video for almost everything. Let’s take a brief look at why livestreaming has been held back so far, and what tech innovations will propel livestreaming to the forefront of internet culture. Right now livestreaming is limited to just a few applications for mass public use and the rest are targeted towards businesses. Livestreaming is to today what home computers were in the early 1980s. The world of livestreaming is waiting for a metaphorical VIC-20, a very popular product that will make live streaming as popular as video through iterations and competition.

Shared Video
Do you remember when YouTube wasn’t the YouTube you know today? In 2005, when Steve Chen, Chad Hurley, and Jawed Karim activated the domain “www.youtube.com" they had a vision. Inspired by the lack of easily accessible video clips online, the creators of YouTube saw a world where people could instantly access videos on the internet without having to download files or search for hours for the right clip. Allegedly inspired by the site “Hot or Not”, YouTube originally began as a dating site (think 80s video dating), but without a large ingress of dating videos, they opted to accept any video submission. And as we all know, that fateful decision changed all of our lives forever. Because of YouTube, the world that YouTube was born in no longer exists. The ability to share videos on the scale permitted by YouTube has brought us closer to the “global village” than I’d wager anyone thought realistically possible. And now with technologies like Starlink, we are moving closer and closer to that eventuality. Although the shared video will never become a legacy technology, before long it will truly have to share the stage with its sibling, livestreaming. Although livestreaming is over 20 years old, it hasn’t gained the incredible worldwide adoption YouTube has. This is largely due to infrastructure issues such as latency, quality, and cost.
Latency is a priority when it comes to livestreams.
Latency is the time it takes for a video to be captured and point a, and viewed at point b. In livestreaming this is done through an encoder-decoder function. Video and audio are captured and turned into code, the code specifies which colours display, when, for how long, and how bright. The code is then sent to the destination, such as a streaming site, where it is decoded into colours and audio again and then displayed on a device like a cell phone. The delay between the image being captured, the code being generated, transmitted, decoded, and played is consistently decreasing. It is now possible to stream content reliably with less than 3 seconds of latency. Sub-second latency is also common and within the next 20 or so years we may witness the last cable broadcast (or perhaps cable will be relegated to the niche market of CB radios, landlines, and AM transmissions).
On average, the latency associated with a cable broadcast is about 6 seconds. This is mainly due to limitations on broadcasts coming from the FCC or another similar organization in the interests of censorship. In terms of real-life, however, a 6 second delay on a broadcast is not that big of a deal. In all honesty a few hours’ delay wouldn’t spell the doom of mankind. But for certain types of broadcasts such as election results or sporting events, latency must be kept at a minimum to maximize the viability of the broadcast.

Sensitive Content is Hard to Monitor
Advances in AI technologies like computer vision have changed the landscape of internet broadcasting. Before too long, algorithms will be better able to prevent sensitive and inappropriate content from being broadcast across the internet on livestreaming platforms. Due to the sheer volume of streams it is much harder to monitor and contain internet broadcasts than it is cable, but we are very near a point where the ability to reliably detect and interrupt inappropriate broadcasts instantaneously. Currently, the majority of content is monitored by humans. And as we’ve learned over the last 50 or so years, computers and machines are much more reliable and consistent than humans could ever be. Everything is moving to an automated space and content moderation is not far behind. We simply don’t have the human resources to monitor every livestream, but with AI we won’t need it.
Video Quality
In the last decade we have seen video quality move from 720p to 1080p to 4K and beyond. I can personally remember a time when 480p was standard and 720p was considered a luxury reserved for only the most well funded YouTube videos. But times have changed and people expect video quality of at least 720p. Live streaming has always had issues meeting the demands of video quality. When watching streams on platforms like Twitch, the video can cut out, lag, drop in quality, and stutter all within about 45 seconds. Of course this isn’t as rampant now as it once was, however, sudden drops in quality will likely be a thorn in the side of live streams for years to come.

Internet Speeds
Perhaps the most common issue one needs to tackle when watching a live stream is their internet speed. Drops in video quality and connection are often due to the quality of the internet connection between the streamer and the viewer. Depending on the location of the parties involved, their distance from the server, and allocated connection speed the stream may experience some errors. And that’s just annoying. Here is a list of the recommended connection speeds for 3 of the most popular streaming applications:
Facebook Live recommends a max bit rate of 4,000 kbps, plus a max audio bit rate of 128 kbps.
YouTube Live recommends a range between 1,500 and 4,000 kbps for video, plus 128 kbps for audio.
Twitch recommends a range between 2,500 and 4,000 kbps for video, plus up to 160 kbps for audio.
Live streams are typically available for those of us with good internet. Every day more people are enjoying high quality speeds provided by fibre optic lines, but it will be a while until these lines can truly penetrate rural and less populated areas. Perhaps when that day comes we will see an upsurge of streaming coming from these areas.

Language Barrier
You can pause and rewind a video if you didn’t understand or hear something, and many video sharing platforms provide the option for subtitles. But you don’t really get that with a live stream. Pausing and rewinding an ongoing stream defeats the purpose of watching a stream. However, the day is soon approaching where we will be able to watch streams, in our own native language with subtitles, even if the streamer speaks something else. Microsoft Azure’s Cognitive Speech Services can give livestreaming platforms an edge in the future as it allows for speech to be automatically translated from language to language. The ability to watch a livestream in real time, with the added benefit of accurate subtitles in one’s own language, will also assist language learners in deciphering spontaneous speech.

--

--