Salsa Sound

An innovative startup born out of extensive research at Salford University, UK. Initially targeting the broadcast audio market, we have developed & productised a set of tools which help automate & augment live sports audio mixing, providing a more engaging & immersive experience for viewers.

Our automated mixing algorithm uses artificial intelligence to analyse feeds from the existing microphones for content that matches predefined acoustic feature templates; it then localizes any detected ‘audio events’ in time & space using triangulation techniques from the available microphones.

The information gained through this process can be used for current broadcast workflows providing a clean on-pitch feed to the sound supervisor or as part of an object-based audio stream where objects are tagged with (e.g. location & duration) metadata.

Salsa therefore offers an enhanced listening experience to every viewer independent of their playout system as improvements are at the production rather than reproduction end. Additionally, as the on-pitch mix is automated, it frees up the sound engineer from ‘chasing the ball’ with console faders & allows them to focus on the more creative elements of creating a great sounding overall mix, it thus ensures consistent deterministic on-pitch mix quality across matches.

Our technology also provides a real-time data stream describing location & timing of all significant sound events (e.g. ball-kicks, whistle-blows). This data provides value in other (non-audio) parts of the broadcast chain, for example it can inform the automation of camera movements, automated graphic overlays, content tagging & rapid highlight & replay generation. We utilise audio information from the microphone feeds to provide added value right through the broadcast chain.