Providing accurate captioning for accessible videos

The Challenge

A Fortune500 company approached TranscribeMe with the desire to transcribe thousands of research calls each month quickly and with high accuracy. As a multinational company, these calls included a wide variety of accents and covered a large range of topics, including both simple and technical content. The audio also frequently contained background and ambient noise. The company faced a challenge in that the transcriptions were required to be delivered within 24 hours with near-perfect accuracy, and they had been unable to find a single vendor that could meet these requirements.

The Solution

TranscribeMe provides a suite of high accuracy human-generated transcription products. Through a global network of over two million registered workers combined with a process of curating teams of workers for specific use cases, TranscribeMe is able to deliver consistently high accuracy transcription efficiently and at low cost. In order to meet the company’s requirements, TranscribeMe looked to organize a workflow that paired automatic speech recognition (ASR) software with a team of transcriptionists. After comparing and benchmarking a number of different ASR providers, TranscribeMe found that Speechmatics provided the best quality output across different verticals and accents. In partnering with Speechmatics, TranscribeMe has been able to create a hybrid workflow that delivers text output consistently above 99% accuracy and is completed within the customers required timeframe. This process has also enabled TranscribeMe to more rapidly scale up the team capabilities and deliver much greater volumes of data while maintaining low costs.