Your browser is not supported. Please update it.

5 May 2022

HESP licensing limits scope, as low latency market hots up

With the rising interest in low latency streaming, facilitated by CMAF and the new versions of HLS and DASH, it is worth checking in on Synamedia’s low latency bet. As vendors get their ducks in order, with some already handling LL-HLS and LL-DASH workflows for customers, there is still room for a new entrant.

The High Efficiency Stream Protocol (HESP) was debuted by its inventor, Theo Technologies, at NAB 2019. In July 2020, Synamedia jumped on board to launch the HESP Alliance, promising a HESP packager from Synamedia to complement Theo Technologies’ HESP player. In October 2021, G-Core Labs integrated HESP into its CDN delivery infrastructure, joining the dots between the video production environment and the end-devices.

Synamedia’s initial involvement appeared set to encourage a raft of big names to sign up to the HESP Alliance. However, this has not come to pass. Synamedia and Theo Technologies are gold members, with G-Core Labs as a silver. EZDRM’s DRM-aaS and Greening of Streaming backer Mainstreaming are also in the silver tier.

Videon and its edgy live encoders are bronze members, but two unknowns round out the member list – Scalstrm, a monitoring and orchestration platform making next-gen CDN claims, and Video Conference Solutions, which does what it says on the tin. Only eight members in nearly two years does not scream progress, and it appears that the licensing terms are to blame. From what Faultline has ascertained, the HESP Alliance is looking to charge a fee to use HESP, which has slowed interest in the protocol. This sets it apart from the likes of LL-HLS and LL-DASH, as well as the CMAF container that underpins the new

approaches, which are all available on an open basis. The obvious comparison point for HESP is HEVC, where royalty terms led to much turmoil that essentially left the industry looking at its AVC predecessor as the better option.

There is an IETF draft for HESP, meaning that interested parties could create their own implementations. The problem is that the HESP Alliance would then seek to agree terms with the user – although we know that CDNs would not be on the hook for the HESP data that flows through their networks.

On the CDN point, there is still a concern among CDN vendors that moving to UDP for WebRTC is not cost effective, due to the need to construct the necessary reflector networks – to redirect traffic to all the viewers. As HESP remains HTTP-based, the CDNs that would carry HESP traffic would not need to do much, and our understanding is that major providers are already tinkering with HESP.

A little digging on the HESP Alliance website eventually unearths the royalty rates. Burying it inside the Resources tab means that most visitors are going to miss the fact that fees are involved. Annual caps of $25 million, with fees for both number of users and the number of hours watched per user, as well as devices, casts quite a wide net for HESP.

HESP is positioned as an alternative to LL-HLS and LL-DASH, as well as a rival to WebRTC. It is billed as an HTTP-based adaptive bitrate (ABR) cross-platform protocol, which can provide end-to-end latencies of as low as 400ms.

Under the hood, it is apparently quite similar to LL-DASH. It has a slightly odd focus on its benefit times for ‘zapping,’ betraying its desire to be used inside linear broadcast systems – and claims a 20x reduction in zapping time, between channels.

As for bandwidth, HESP claims that it is not focused on using shorter segments, like the current approaches used in LL-HLS and LL-DASH. Consequently, HESP claims a 10% streaming bandwidth reduction, while still achieving sub-second latency.

Additionally, if your target is not low latency, there are other HESP modes that will allow for greater encoding latency, which can increase the bandwidth savings to 20%.

For Synamedia, its support for HESP currently sets it out from the pack. Its online video platform (OVP) positioning means that if a customer specifically wants to use HESP, then Synamedia is currently the only obvious option.

It would not surprise us to learn that other OVPs have explored HESP, but HESP functionality would also need to be added to all the video players to enable the client devices.

And so, we return to the classic video technology question of standards. A larger philosophical point is whether there could ever actually be a single agreed upon way of streaming video?

The closest we have got are the underlying internet protocols, which are foundational to OTT video. Without the internet infrastructure that powers the web, our modern economies would not have emerged – never mind Netflix or YouTube.

At nearly any point higher up the stack than the foundational internet protocols, there are a plethora of overlapping technological options available. In Faultline’s world, codecs are perhaps the best example, but you can examine any point in the video delivery chain and multiple options that do essentially the same thing.

To this end, it seems inevitable that there will be overlaps, and so the emergence of HESP should not draw tears from the LL-HLS and LL-DASH crowd, nor the CMAF supporters.

Different approaches can coexist quite happily, and worse, getting customers to care that there are multiple ways to deliver video is unlikely to be productive.

Most customers only care about price, and whether their purchasing choices fit a given criteria. If the choice is operating as expected, and has cost as much as the budget afforded, then no heads will roll, and everyone will be happy. Evangelizing about the technological purity of your given standard or protocol is simply a waste of breath, as the end-users do not care.

Still, there is an efficiency argument about wasted collective resources, even though it can usually be countered by pointing out vested commercial agreements. Laugh at AOMedia all you like, but it did appear to genuinely want to create a better and free codec than the industry had decided was good enough. Imagine how polished a codec we might have if the dozens of different initiatives could actually collaborate on a single version.