We are proud to announce that our transcription process now uses high quality automatic transcriptions provided by Raytheon BBN Technologies using BBN’s Speech-To-Text engine. This milestone is the culmination of an exhaustive search for the highest quality provider of automatic transcriptions produced using the best available automatic speech recognition (ASR) technology to incorporate into our manual transcription.
The relationship with BBN provides a platform for continual improvement of the draft transcripts, which are subsequently cleaned up by our workforce of more than 250 professional transcriptionists. A crucial benefit of BBN’s ASR engine is its very rich output capabilities. Instead of just a single “best path” draft transcript, BBN automatic transcription provides lattices, search parameters, and information from multiple passes. With this information, we are able to improve the formatting of drafts and to more accurately assess the editing difficulty for a given media. These improved drafts have already made a significant impact, allowing our editors to focus on the finer details of transcription, such as proper punctuation, numerals, formulaic expressions, and correct spelling of names, places, and neologisms.
BBN’s ASR engine is widely recognized as the leading technology for batch speech-to-text processing in the most demanding applications, including some mission critical government deployments. In addition, BBN’s algorithms consistently rate at or near the top in the frequent speech technology evaluations conducted by the National Institutes of Standards and Technology. BBN continues to invest heavily in R&D to improve its ASR technology. Importantly, this relationship includes an ongoing commitment to continually adapt and improve the acoustic and language models, increase the ASR vocabulary, and pursue other improvements to provide consistently high quality transcriptions.
Our relationship with BBN exemplifies our dedication to continually improving our products and services. We expect to make more such announcements in the coming months, as our team is working on many different aspects of the application, with an eye toward improving the quality of our time-synchronized transcripts and closed captions.
