Automating the listening process
Envision Racing wanted a model to sort relevant clips and then group them by driver and race, enabling engineers to act on conversations faster. But because clips are short and often contain encoded messages, standard algorithms can struggle with them.
The first step was to build a transcription platform that could translate speech to text and automatically eliminate dead air. During a full race session, drivers and their engineers typically speak for less than 10 minutes in total. The team uses this part of the solution in real time.
Genpact designed the Radio Analytics Engine (RAE), a platform that quickly processes each recorded radio stream into relevant clips. Now, nobody has to trim the silence from the audio, and the team can quickly catalog clips by driver and race. This makes it easier to gain quick insights on how rivals approach challenges, including when they may take to the pit, how they manage remaining energy, and when they use attack mode – an extra boost of energy available to all drivers during the race.
By introducing natural language processing to RAE, the team will automate clip sorting quickly enough to access insights and intelligence during a race. With more detailed competitive knowledge, drivers can make better in-race decisions, improve overtaking and defensive strategies, and change the outcome of a race.