SEGMENTS:2025 Montreal Agenda
SEGMENTS:2025 Montreal
Speaker Schedule
-
2025-10-31Day 1
MOQtail: a Little Bit Live, a Little Bit On Demand, a Whole Lotta QUIC
MoQ is redefining how live and on-demand media are delivered over the Internet, offering a unified, low-latency and scalable transport layer. In this talk, we present the new features of MOQtail, showcasing the flexibility of MoQ in supporting both real-time and time-shifted content. A key highlight is our implementation of FETCH, a powerful feature designed to optimize on-demand media retrieval. We will demonstrate FETCH live at IBC 2025, illustrating its potential to simplify infrastructure and enhance user experience.
How to replace specific ads in a break in Linear TV?
Linear TV still commands the most ad impressions—but delivering addressable campaigns on these premium assets requires surgical precision. This session introduces how to utilize SSAI, SCTE-35 Time Signal, and a few other methodologies in the ecosystem to enable broadcasters and TV distributors to replace only the most irrelevant ad spots within breaks, while preserving the viewer experience and unlocking new digital inventory. Using TF1+ and RMC-BFM as examples, we’ll walk through the technical methods, business value (reach, efficiency, and measurement), and operational advantages (QoE preservation and load smoothing).
Mathias Guille
Building an Open Intercom for Live Production: Lessons from the Nordic Broadcasters
Nordic broadcasters faced a clear challenge: traditional intercom systems for live production were closed, expensive, and difficult to integrate with other broadcast workflows. To solve this, we set out—together with several broadcasters—to design and implement an open, cloud-native alternative. The collaboration was essential: broadcasters shared their practical requirements from sports, news, and entertainment production, and we transformed those into a working architecture. By leveraging WebRTC and open audio routing standards, we achieved low latency, device interoperability, and seamless integration with existing systems. The result is a production-ready, open-source intercom now running in broadcaster environments. It reduces costs, avoids vendor lock-in, and demonstrates how cooperation between technology providers and media organizations can deliver real, usable solutions to long-standing production problems.
Magnus Svensson
Content Delivery at 30,000 Feet: Content Streaming Challenges To A Connected Aircraft
Inflight entertainment (IFE) is undergoing rapid technological change driven by the rise of high-speed connectivity. With the introduction of Starlink and competing offerings from Amazon, Viasat, and others, inflight connectivity is becoming increasingly ubiquitous. Connected aircraft open the door to new possibilities such as streaming services, enabling the adoption of technologies like edge caching and casting—areas that have been the focus of several SVTA workgroups. However, implementing these technologies in IFE systems presents unique challenges, and the migration from ground-based to airborne environments is far from straightforward.
Juraj Siska
Break
Container-Based Delivery Platform
This session explores how Viasat partnered with 2you.io to deploy a container-based, multi-tenant CDN platform built on distributed Kubernetes. We’ll cover how Open Caching integration, CI/CD alignment, built-in observability, and edge application capabilities enabled Viasat to scale globally, accelerate deployment, and deliver customized content workflows—now live in two regions with further expansion planned.
App Meets Transport: Rethinking Video Streaming over QUIC
The migration to QUIC often fails to deliver the promised video QoE gains. This presentation argues that this is not a failure of QUIC itself, but of the siloed paradigm where the application, transport, and network layers operate blindly to one another. Using evidence from recent research, we diagnose this fundamental "information gap" and present a new architectural model for a symbiotic streaming stack. We will showcase data from case studies demonstrating that a cross-layer approach—where the app, transport, and network layers cooperate—delivers dramatic QoE improvements, including over 80% increases in VMAF and 30% reductions in latency.
Abdelhak Bentaleb
Best Practices: Development of an AI Agent Connected to an MCP Server
In our company, we have a product for primary distribution. Simple and nicely polished User Interface (UI), which leverages a well-documented API. Though, it is a click-fest. We would like to explore the benefits of using natural language interface, so we developed a Model Context Protocol (MCP) server to deal with the API, and we implemented an LLM agent to chat with the administrator and play with the Product. During the presentation, we will explain how it works, its Use Cases, what we learned from this development, and the effect on our industry.
Gwendal Simon
ASICs in the Era of Modern Communications: A Deep Dive
As video streaming, gaming, surveillance, and real-time communications continue to scale, traditional compute architectures are reaching their limits. This session explores how Application-Specific Integrated Circuits (ASICs) are transforming high-bandwidth media workflows by delivering ultra-efficient, high-density, and low-latency video processing. We’ll dive into the architectural benefits of ASICs over CPUs, GPUs, and FPGAs, and show how purpose-built hardware accelerators enable more scalable, energy-efficient infrastructure across applications like live streaming, VOD, CDNs, cloud gaming, gambling platforms, and AI-powered surveillance. Attendees will gain technical insights into ASIC design, codec support (H.264, HEVC, AV1), software integration with FFmpeg and GStreamer, and real-world deployment benchmarks. We’ll also explore how ASICs reduce rack space, power draw, and cost while opening new opportunities for infrastructure growth and operational savings. Whether you're an engineer or decision-maker, this talk delivers practical value and a forward-looking perspective on how ASICs are becoming essential in modern communication workflows.
Leo Nieto
Fitting Horizontal pegs into Vertical Holes: Reformatting Video for Phones
While the preferred format for short-form social media video consumption on phones is vertical video, the long-form video universe has been slow to embrace this format. Our company is creating a mobile-first video podcasting platform, where vertically formatted video presents the podcast speakers up close in a way that feels more intimate - more like a one-on-one video chat. Our challenge is that virtually all video podcasts are shot and edited for the preferred Youtube format - horizontal. We have developed a novel approach leveraging machine vision, speaker identification, unsupervised learning, and custom algorithm development to automatically reformat horizontal video to vertical video, keeping the focus on the person speaking.
Lunch
Panel: AI, What\\\'s it Good for?
Bhavesh Upadhyaya
Barry Owen
Magnus Svensson
Media Quality Assessment (MQA) metadata standard: new tool for monitoring and optimizing streaming QOE
The talk will give an overview of Media Quality Assessment (MQA) metadata standard, currently under development by the SVTA QOE working group. We will explain the motivations for this standard, its relation to CMSD and CMCD metadata, and its use cases and utilities. As an example, we will explain the use of this standard for more complete and accurate assessment and reporting of QOE delivered by streaming media systems on different categories of devices (mobiles, TVs, PCs, tablets, etc.).
Break
Broadcast Production over Private 5G
As live broadcast workflows evolve, the industry is turning to private 5G networks to enable new levels of agility, scalability, and efficiency. In this session, we will explore how the use of private 5G networks is transforming live broadcast productions and coverage of dynamic sports, news, and events. Through real-world examples, including deployments at major global sporting events and the IBC2025 Accelerator project with France Télévisions, Ghislain will demonstrate how new live video contribution technologies are empowering broadcasters to deliver high-quality, low-latency content securely and reliably. These innovations are enabling media organizations to unleash creativity and stay close to the action.
Ghislain Collette
AI + Metadata & QC: Preventing Revenue Leakage in Your Streaming Tech Stack
Product-led strategies for metadata integrity, QC automation, and revenue assurance.This talk shows how AI in metadata and QC prevents revenue leakage across FAST, AVOD, and broadcast-style channels. We begin with the North Star for QC—reliability, compliance, viewer trust, and ad integrity—and why upstream correctness (accurate metadata, frame-accurate SCTE-35/104 markers, and signal-preserving encoding) is foundational. We then deep-dive a FAST channel workflow where continuous monitoring, AI/ML detection, and a business-rules layer automatically correct missing/late cues, mis-tags, and incomplete ad fills. Real-world product examples ground the story: Bitmovin’s AI Scene Analysis adds scene-level metadata that boosts discovery and contextual ad placement, with IBC 2025 updates for highlights and summarization; Interra’s BATON 9.3 expands AI-powered QC (blank-bar and color-gamut checks, PSE alignment, ASR/NLP captions). A STIRR case shows scene intelligence at scale (≈30k titles/300 channels) improving operations and monetization. We bridge to in-house fixes with three DIY patterns (synthetic SCTE insertion, safe LLM auto-patch for ratings/tags, scene-level IAB/brand-safety) and a version-pinned regression harness so LLM/API changes don’t create new leaks. Attendees leave with a practical roadmap—audit, pilot, version, test, monitor—and a mini calculator to quantify revenue at risk.
Bhavesh Upadhyaya
Flash Talks