OTT live sports has led sports broadcasting to a new level. Yet, there are still some challenges. For example, nobody wants to hear a neighbor’s cheers when a goal is scored before seeing it on the screen, making low-latency transport and playback indispensable. Synchronization among all viewing devices and social media feeds is also essential.
The IETF recently started developing a QUIC-based low-latency delivery solution for media ingest and distribution. The Media-over-QUIC work is still in its infancy but could be a game-changer. In this talk, we will discuss the architectural design issues and present preliminary results from our implementation.
Third Room is an open and decentralised vision of the metaverse for the open Web, built entirely on the Matrix open standard… there’s no cryptocurrencies, NFTs or walled gardens. Make it yours by building new worlds or adding interactive objects in the existing ones! Also our avatars have legs.
WebRTC is often considered to be secure by default - with most security concerns being around IP address leakage which is more of a privacy issue than anything. Well, I have news for you - the applications and infrastructure that handles WebRTC can be attacked. It may indeed have various types of security vulnerabilities which are often overlooked. This presentation is based on experiences gained through security testing of WebRTC applications with anecdotal stories to illustrate the dangers. We will also take a peek at Video Delivery mechanisms such as RIST and SRT and discuss what could possibly go wrong there too!
This talk gives an overview about the new features and improvements in the upcoming Kamailio 5.7.0 release series. Kamailio is one of the leading Open-Source VoIP servers, powering many RTC infrastructures world-wide. The upcoming release 5.7.0 introduces several new modules. It comes with many improvement to existing important modules and many other fixes in code and documentation. Furthermore important changes related to deployment, packaging and containerization will be presented.
In this talk, I'll discuss how modern technology has blurred the lines between traditional real-time communications (phone calls!) and traditional broadcast media (video!)
ML based voice interfaces have been kicking around in hype cycle for 6 or 7 years. Its is a painstaking task to build a decent one, and, lets face it users have stopped expecting much so it is hard to get excited about the return on effort needed to build them. This talk implements one live onstage in 20mins with a GPT rather than hand crafted dialog training and tries to figure out how much of a game-changer this approach will be.
Some small, cheap, light linux boards have hardware video codecs that can produce 1080p @30pfs h264 at very low power consumption - making them idea for streaming low latency content from moving sources such as droids and vehicles
In this talk, we will explore the exciting field of video analysis and examine the use of computer vision and data processing techniques. Through real-life examples and case studies, we will showcase the latest algorithms used in this area and discuss their potential applications. In addition, we will provide an in-depth look at a recent project developed by the data team at Noumena. This innovative project centers on traffic monitoring using video and artificial intelligence in the vibrant city of Barcelona.
Telecoms used to be obsessive about testing - but that's a thing of the past. The growing complexity of realtime communication services, paired with the grinding halt of the evolution of relevant test tools, leaves RTC developers and operators in the dry. In this talk, we'll show some modern approaches to automated telecom testing, using open source tools you already know and love.
The Fedora Design Team hosted the very first Creative Freedom Summit in January 2023. The 3-day event was a wild success, with roughly 635 unique attendees total from 54 countries and up to 265 concurrent live viewers. We used a completely free & open source technology stack, relying on Matrix's element.io service as the central conference home, streaming presentations from Jitsi live to Matrix via PeerTube. In this talk, you will learn how we set up the tech stack for this event, how it went day-to-day for the small volunteer team running it, and our lessons learned for next year's event.
Events and meetings have changed a lot in the past few years. What typically envisaged people meeting and engaging in some location around the world (possibly with a handful of active remote participants), first changed to a fully virtual participation, and is now moving towards a more seamless interaction between many on-site participants, and possibly as many remote participants instead. This presentation will explain how Meetecho handled this (double!) transition, and will provide some information on how it helped actual hybrid events (e.g., IETF and RIPE) finally take place for the first time with the help of ad-hoc tools (and Janus!).
Cross browser compatibility got a lot better in recent years. There is a huge set of browser APIs which work very very similar in all modern browsers. The implementations are so close that many of the CLIs that come with todays frontend frameworks set up tests that only run in one browser or no browser at all.
But unfortunately things are still very different when it comes to media APIs. There are a lot of APIs which are still new and work in progress. They often don't work as promised by the specifications and they often work differently from browser to browser or they don't work at all. It's not uncommon to run into a browser bug.
Reporting a bug is the mandatory first step to get it fixed. And therefore it's best to have a reproducible test case which demonstrates the bug. Creating this test case is often the tricky part. But once it's done it can also be turned into a Web Platform Test.
Web Platform Tests are a set of tests that verify browser implementations. All big browser vendors got together to create a large corpus of tests that they run across all browsers. The results are used often to show off spec compatibility. Web Platform Tests are for example used by the yearly Interop projects. But it's an open source project. Anyone can contribute a test.
Adding your test ensures visibility. It will lower the compatibility score if it fails. It will probably also increase the motivation to get it fixed. And last but not least it will make sure that the same bug never pops up again in any of the tested browsers.
Let's improve browser compatibility together one test at a time.
Many people believe that WebRTC is the only option regarding sub-second latency streaming due to its low latency and widespread browser support. However, other options may be better suited for specific use cases, especially when considering feedback from the open-source community and video cloud service customers.
Depending on the specific use case, WebRTC may be the best choice for web browsers, while SRT may be better suited for broadcasting devices. However, combining both may be necessary in some cases, and RTMP plus WebRTC can also provide sub-second latency.
In this discussion, I will explore the most effective solutions for these use cases using SRS technology to provide sub-second latency streaming services. Additionally, I will introduce the SRS project and community, which has existed for ten years and has over 20,000 stars on GitHub, making it a valuable resource for developers worldwide.
This talk will provide an in-depth analysis of anycast routing for media, specifically focusing on protocols used by Cloudflare Stream and Cloudflare Calls, such as RTMP, SRT, and WebRTC. Anycast routing offers numerous advantages like reduced latency, increased reliability, and load balancing. However, it also introduces challenges. The discussion aims to help understand how Cloudflare effectively employs anycast routing in the real world.
In this talk, I will showcase a live demo of a cutting-edge system, code named S.A.T.U.R.D.A.Y, that uses your voice to interface with an AI-powered personal assistant just like Tony Stark's J.A.R.V.I.S. We will connect to a WebRTC call and demonstrate how the system can seamlessly translate audio to text, pass that text to an AI model, and then translate the text output of that model back into audio in real-time, all while remaining connected to the original call.
My demo will highlight the potential of this system for a wide range of applications, from personal assistants to language translation, and even voice-activated devices. I will also explore the current state-of-the-art in voice recognition and AI, and discuss how new advancements in these technologies are opening up new possibilities for innovation and creativity.
Whether you are a developer interested in AI and voice technology or simply curious about the cutting-edge of human-machine interaction, this talk is for you. Join me as we explore the exciting world of Tony Stark in real life, and discover the possibilities of voice and AI.
The Matrix ecosystem is well known for applications such as Instant Messaging, but what if you want to transmit real-time data such as video conferences or sensor data? MatrixRTC is the world's first decentralised low-delay platform to exchange real-time data between groups of people over Matrix. MatrixRTC enables a new set of applications which benefit from low delay and real-time properties.
In this talk we walk through the journey from 1:1 calling towards the powerful MatrixRTC framework which powers Element Call, our video conferencing solution.
Whether you are familiar with jambonz or not, if you are interested in building voicebots or any kind of programmable voice application using jambonz open source technology this class is for you. We will quickly get you up and running and then move on to more advanced topics. You will get a thorough overview of the jambonz technical architecture and learn how to build applications using either both webhooks and wesockets. We'll also cover a bit of history about the jambonz project as well as future roadmap ideas and opportunities to participate in the community.
This workshop will address the challenges faced by web developers when working with tracks acquired from the getUserMedia API, including permission prompts, cross-browser compatibility, dealing with virtual devices, managing track states and solving auto playback of media. We will provide solutions for optimizing the acquisition, retention, and playback of these tracks and solving all the aforementioned problems.
By the end of this workshop, you'll be equipped with the tools and knowledge to tackle any getUserMedia API track challenge and have a fully functional, cross browser, cross device compatible working GUM ready app, Don't miss out!
Open source software has hit the mainstream with an unprecedented trajectory which has not been matched by a growth in understanding. Whilst many would say that the war has been won and today open source is mainstream and the majority of software developed, it faces many challenges to a successful future. The question of course is will open source fail?
Everybody lies. In this nasty world of end-to-end video delivery, all the involved stakeholders lie to achieve their own results. But where is the truth? The truth lies right in front of your eyes, and it’s always been there - in your CDN logs.
Having spent almost a decade in this industry, working my fingers to the bone with FFMpeg commands, and ending up crying drunk a few times in Vegas & Amsterdam being agonised over high latency, glitches, and stalled video errors…my rocky journey made me learn a few things about CDNs, and now it's time to share my industry expertise in CommCon.
During this session, I'll share some tips on understanding the logs' info, ways of predicting some bad behaviours in your CDN network, and how to fix the issues before our viewers face the dreaded 400s.
Follow me on this logging journey where you'll discover some insider secrets of the video industry, and most importantly learn how to predict stream issues thanks to our wonderful CDN logs. Don't trust anybody except your headers!
Open5GS is a C-language Open Source implementation for 5G Core and EPC, i.e. the core network of NR/LTE network. (Release-17). This presentation explains how to build your own 5G and LTE Networks with Open5GS.
The Bloomberg Terminal is used by more than 350,000 finance professionals around the globe. To help them with their communications and compliance needs, Bloomberg's engineers have used open source software to build a variety of useful RTC workflows within the Terminal. In this talk, we will cover some of these workflows and how WebRTC has enabled us to implement them.
React Native has been powering Jitsi Meet and countless other applications for many years now. Let's take a look under the hood and see how we made its biggest change yet: the migration to Unified Plan, amongst many other WebRTC war stories.
WebRTC may boast broad support across browsers and clients, but anyone who's tried to build with it knows that consistency is not its strong suit. Over the past couple of years, we've navigated the murky waters of WebRTC quirks, sometimes hair-pullingly frustrating, always enlightening.
Join us as we dive into the world of WebRTC oddities, exploring the differences across browsers and mobile clients. We'll discuss everything from codec support and audio handling to API variations and those pesky undocumented behaviors that catch you off guard.
Whether you're a seasoned WebRTC developer or just dipping your toes into the pool, this talk will provide valuable insights and leave you with a deeper understanding of how to make WebRTC work for everyone, everywhere.
SIPssert is an open-source testing framework that aims to ease VoIP scenario testing - this talk is about what it offers and how it can be leveraged to provide conformity tests for your VoIP platforms.
Frequently, developers overestimate humans. Particularly when making design choices. This can be the big mistake that dooms a project… We’re going to discuss well-invested engineering efforts like benchmarking and optimizing minimalistic data flows. When is a user interface too flexible...? We’ll also talk about how to avoid tragedies like giving users (or their devices) too much credit.
By leveraging Matrix as the engine to carry RTC and text traffic we built an omnichannel communication component in our Unified Communications as a Service product for occasional business users.
As opposed to call-centre centric software, we wanted to offer a simple centrally integrated component for our SPA to give Cloud PBX telephony users the ability to engage with clients from a unified platform where their tools are all available in one interface - Phone, Video Meetings, Chat, Whatsapp Integration, website chat - all enriched with data from the user's CRM.
