Live sports are notoriously difficult to mix and require a lot of cognitive attention in order to achieve good reults. The goal as always is great storytelling, but difficulties in managing the audio mix for live sports mean that not all of the necessary sounds of the game are included. This leaves the content less engaging, and viewers lose some of the match narrative.
In this talk we will discuss how we have approached this problem using machine learning to listen for sounds with specific acoustic features, and to add these salient sounds automatically to the mix. The aim is to produce a more engaging, enhanced audio mix which better tells the story of the match or game. We will describe this approach in more detail, including how we are deploying this technology in the cloud for distributed remote production using 5G technology and edge-compute capabilities.
As well as the on-pitch sounds that provide the match narrative, the crowd sound provides emotion and is vital for a great story. We will describe how we have emphasised/enhanced this to produce a ‘fan experience’ mix. Our intuitive virtual crowd app has been used throughout the world to create dynamic, realistic virtual crowd sound for games played behind closed doors during the pandemic.
Rob Oldfield completed his PhD at the University of Salford in 3D Audio technology, and continued working at the university in research and consultancy for several years. Rob’s research interests are primarily in improving broadcast audio and developing new audio capture technologies for a better end-user experience. Patented technology from this research led to the recent formation of spin-out company Salsa Sound and received backing from the Royal Academy of Engineering, which appointed Rob as one of its Enterprise Fellows. Salsa Sound focuses specifically on sports broadcast and Rob’s goal is to make the sound of sport on TV more engaging, more cinematic and more interactive.
This online lecture is open to all. Registration is required via Zoom (Link to be posted soon). Please note this is just for registration: a separate link to the meeting will be sent to registered participants prior to the talk.