University of Surrey researchers build AI tool to improve Supreme Court transcriptions

Reading time: one minute

Researchers at the University of Surrey have built an AI tool that transcribes Supreme Court hearings and “makes justice more transparent and accessible”, according to a new study. The tool, trained on 139 hours of courtroom audio and legal documents, is said to reduce transcription errors by up to 9% compared to leading commercial systems.

The custom speech recognition system was specifically tailored to the unique language used in British courtrooms, making it more effective in capturing legal discourse. It tackles legal terminology errors that generic tools often misinterpret, such as mistaking “my lady” for “melody” or “inherent vice” for “in your advice”. The tool is also said to be more reliable at picking up provisions, case names and judicial titles. 

Professor at the University of Surrey and co-author of the study, Constantin Orăsan, said: “Our courts deal with some of the most important questions in society. Yet the way we record and access those hearings is stuck in the past. By tailoring AI to the unique language of British courtrooms, we’ve built a tool that makes justice more transparent and accessible – whether you’re a barrister preparing an appeal or a member of the public trying to understand why a judgement was reached.” 

In addition to transcription, the AI system semantically links written judgements to the exact moment in hearing videos where arguments were made. A prototype interface allows users to click on a paragraph in a judgement and instantly view the corresponding courtroom exchange. User testing revealed a significant productivity boost: legal experts using the AI tool validated 220 links in just three hours, compared to 15 hours needed to identify 10 links without AI assistance.

The AI tool is already drawing interest from the UK Supreme Court and the National Archives, according to the University of Surrey.