The talk will cover what goes into delivering a reliable and high quality audio experience that scales across the diversity of device and network conditions of our users across Meta's family of apps including Messenger, Instagram, Facebook and WhatsApp. In addition to covering components such as gain control, echo cancellation, noise suppression and codecs that are typically spoken about, the talk will dive deep into a less familiar but arguably more important problem of audio reliability (can you hear me?). The talk will also address new challenges in delivering audio for the Metaverse where users experience the magic of being present in the same room.

AV1 is becoming mainstream in the streaming world, and now is the time to leverage it for real-time communications as well. Upgrading to a next generation video codec possesses a number of challenges and opportunities when serving over a billion users. Niklas will share Meta's learnings from AV1 integration, discuss tradeoffs and mitigations that can be taken to reduce the negative impacts, and cover the biggest gain that can be made.

Jonathan will describe his successes and failures over his career in Real-time Communications (RTC), and what he believes is core to building, deploying and maintaining successful services that delight users.

Moderated by Li-Tal Mashiach

Enabling RTC from anywhere is becoming increasingly important as we continue to evolve the hybrid workspace and extend the capabilities of RTC. Android appliances address a number of use cases that go beyond what PCs and mobile phones can provide but with that come a host of challenges that need to be addressed. In this session, you will gain a broad understanding of some of the considerations in developing RTC solutions on appliances other than PCs and mobile phones.

Intro to the challenges we are facing to support large group calls. And how our technologies evolve to support large group calls, including BWA, simulcast, temporal layer etc with two audio features for group call: audio capping and dominant speaker detection.

Gain insights into the incredible scale of WhatsApp's calling feature, which handles billions of calls every day. Discover how our calling relay reliably manages such high loads and learn about our strategies to prevent catastrophic failure during extreme traffic spikes. Explore some of the complex scenarios that the calling relay handles and how our design decisions prioritize both privacy and scalability.

Hosted by Deniz Hastorun

Dolby.io is an API platform to enable easy integration of high-fidelity audio, video solutions into your real-time communications, media processing and streaming applications in browsers and mobile applications. In this talk we will walk through the challenges, learnings and approaches taken to enable testing the platform at scale. We will explore how we use a custom DSL together with known architecture patterns like the publish subscribe model to elevate our scale testing and analysis capabilities for video, audio and network parameters at scale

Real-time visual communications are popular nowadays. Object assessment of transmitted video quality, which is correlated with the human visual experience, is challenging in two aspects. First, there is no reference video. Second, end terminals have limited computational power. A machine-learning-based solution to blind quality assessment of real-time video to be implemented in end terminals is presented in this talk.

Meta uses multiple levels of simulations to help improve user experience and prevent regressions before changes impact users. Learn what types of testing work best for various areas of the stack. Find out how to “shift left” and save time and money by gaining learnings earlier in your process. Iterate faster without causing user pain. See how we can test calls with thousands of users… without needing thousands of users.

The Amazon Chime SDK enables builders to add communication capabilities to their apps. A key capability of the SDK is the ability to apply machine learning to enrich audio and video sessions. Customers also use machine learning to derive intelligent insights from these sessions. We will dive deep into how customers are using real-time machine learning capabilities in their communication services.

Moderated by Li-Tal Mashiach

A smaller, faster video calling library for our apps

Discover how Meta is transforming traditional phone calls into fully immersive avatar experiences, allowing you to be present in any virtual environment. We'll explore the cutting-edge technology behind real-time state management, giving you a glimpse into the complexity of creating RTC as an operating system-level construct.You'll also learn about the various types of avatars and how we transport their skeletal data using WebRTC.

Moderated by Deniz Hastorun