Verbit’s proprietary adaptive ASR technology is based on 3 models: linguistic, acoustic and contextual events and is specifically trained for the domain of the customer. This produces a highly accurate initial transcript (average of 90%) just minutes after the file is submitted.
Once the initial technology-produced transcript is ready, the file passes to our human transcribers. This step is divided into two stages: editors and reviewers to ensure that the transcript reaches over 99% accuracy. All our transcribers are highly qualified and must pass a rigorous test with a 10% acceptance rate.