Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now

×

The challenge of low-latency

Stefan Lederer, CEO and co-founder at Bitmovin, looks at how live low-latency streaming remains a challenge, and how AV1 could be 2022's breakout star

Video is expected to account for 82 per cent of all internet traffic by the end of 2022, up from 67 per cent in 2016. Unsurprisingly, during the Covid-19 pandemic, the video streaming industry has seen substantial growth rates in consumption and accelerated innovations across the entire ecosystem – from personalisation to video quality optimisation. 

As the world continues to emerge from the pandemic, this too will impact the video streaming industry and present new opportunities and challenges for video developers who are constantly trying to meet the demand for improved viewing experiences without causing costs to spiral out of control. In a bid to get a snapshot of the current state of play in the video streaming sector, Bitmovin recently surveyed 538 video developers, industry experts, customers, and prospects from 65 countries for our 5th Annual Video Developer Report. We found that this nascent industry is continuing to mature across the board. 

Live low-latency streaming remains a challenge

In 2020, the biggest challenge for video developers was cost. Yet, while controlling the costs of bandwidth and storage remains high on the list of challenges for video developers, low latency topped the list as the biggest headache for video developers, which isn’t entirely surprising. Despite significant technological advancements, live low latency is still a conundrum for the video streaming industry and lags significantly behind broadcast. Most respondents expect low latency video in the 3 to the 5-second range, while 25 per cent of our respondents expect sub-second delays for the second straight year in a row. However, many of the necessary measures streaming services take to scale up and improve reliability amid rising demand end up increasing their end-to-end latency. Solving the issue of live low latency requires examining and optimising several points in the delivery path from encoding, packaging and DRM to the player and CDN strategy. There are trade-offs between quality, cost and scalability when trying to reduce latency, so each service needs to find the approach that delivers the most value for their viewers and use case.  

AV1 is primed for the spotlight 

We have always been champions of the AV1 codec and its potential but we also acknowledge that it’s a multi-codec world. However, the results from the Video Developer Report indicate that AV1 will experience a significant uptick in usage. Currently, AVC/H.264 is the most used video codec because it has the greatest device reach. Companies will need to continue supporting H.264 to ensure interoperability with every device. However, our report found that HEVC, VP9, and AV1 saw significant increases in testing and production usage, with HEVC being deployed by over half of the respondents. The ecosystem may be ready to make the jump to AV1, considering AV1 adoption by consumer platforms like YouTube and increasing device support from the Fire TV Stick 4K Max. Qualcomm also recently announced its support of AV1. Our report found that 22 per cent of video developers plan to use AV1 over the next 12 months, suggesting it could be this year’s breakout star. 

New priorities for AI and machine learning 

In 2020, ‘recommendations and personalisation’ topped the list as the artificial intelligence and machine learning features that would improve the video experience for viewers. However, this year they were surpassed by audio transcription and speech-to-text and ‘Tagging and categorising video content’. A total of 43 per cent of respondents said they were using AI and ML in their audio transcription and speech-to-text efforts. There are two significant cases for using video transcripts. The first is providing closed captions for video content, an essential part of any video strategy. Facebook, for example, discovered that an enormous 85 per cent of video content on their platform was watched on mute. Therefore, closed captions are crucial to providing context for this growing number of users watching content with no sound. A less obvious benefit for transcripts is the increasing discoverability of video assets. Making content easy to find becomes increasingly essential as video archives continue to grow, leading us to the second use case, which is tagging and categorising video content where the benefits lie in storage, search, and sharing videos

The video streaming industry remains one of the most exciting technology industries. It only started taking off around 20 years ago, and the pace at which it’s continued to evolve is astonishing. Over the next 12 months, we expect increased usage of newer and more efficient codecs. For media companies, having access to wider codec options is a sound way to optimise the high-quality viewing experience for viewers. Additionally, increasingly widespread adoption of AI and ML will continue to improve the viewing experience for audiences.