Speech Recognition

The AutoScript™ speech recognition product produces accurate and fully formatted first draft transcriptions from clinicians' dictations. The goal of these drafts is to minimize editing time by MTs and, as a result, increase MT productivity. AutoScript is cutting edge speech recognition technology and so much more. AutoScript interprets and formats a document, so that it is close to a final product. This advanced speech understanding process is a key component of eScription's approach to Computer Aided Medical Transcription (CAMT), and an important reason why eScription is leading the market in CAMT today.

eScription owns its speech recognition engine and controls all new enhancements without relying on another vendor. Our unique speech processing technology was designed by our team and built from the ground up to convert medical dictation to high quality patient documents. Other technology providers use commercial, real-time recognizers that simply are not capable of handling the depth and breadth of medical vocabulary and customized formatting that health care organizations demand. Other speech engines require explicit training by clinicians, forcing them to speak in a certain way. Not so with AutoScript.

	Organized and formatted document sections
	Punctuation inserted even if not spoken
	Numbers interpreted and presented appropriately. This includes dosages, measurements, lists, etc.
	Formatting based on each organization’s preferences and specifications
	Inserts speech-activated ‘normals’
	No explicit training required
	Continually learns and improved from MT edits

Technology
Our AutoScript technology creates high quality first draft documents with the primary goal of reducing the amount of editing required and thereby streamlining the medical transcription process. We do this through:

	We use sophisticated contextual information at the document, sentence and phoneme level to produce accurate content and correct formatting.
	AutoScript rapidly processes each dictation multiple times; this allows the software to adapt to the acoustic environment (e.g., type of phone), and to use the entire document to resolve any ambiguities throughout.

	We create language and formatting models specific to every clinician and to the work type being dictated.
	We build models from billions of words of medical dictations, and this database continues to grow.
	AutoScript continually updates its models, tracking changes in the speaker's voice, the communication channel, and even the dictation and transcription style.

	We incorporate specified formatting rules from each health care organization.
	AutoScript learns from transcriptionists' corrections, adapting the speech recognition models continually.
	AutoScript supports 'voice macros' and can integrate clinicians' 'normals' through spoken instructions.

Process
During the first phase of a new system implementation, clinicians' dictations continue to be typed from scratch while AutoScript works 'behind the scenes' to build clinicians' speech models. Only when it is clear that the AutoScript drafts can improve the productivity of MTs, do the MTs start receiving these drafts instead of typing the dictations from scratch.

Architecture
AutoScript is deployed in a secure, server-based environment. This architecture enables eScription to continually add to our database of medical transcriptions and to continually update our models with more data. Current customers and new customers alike benefit from technology that is enhanced and improved on an ongoing basis through practical application and real-world interactions.

Back to top