Is voice recognition ready for prime time?
By ARTHUR GASCH
Healthcare InfoTech Contributing Editor
Voice recognition was a big topic at the Healthcare Information and Management Systems Society (HIMSS; Chicago) conference in Atlanta last month, not necessarily because a lot had changed on the technology front, but rather, because a lot had changed in vendor commitment.
While IBM (Armonk, NY), Dragon Systems (Waltham, MA), Philips Business Systems (The Netherlands), and others were present, it was Gaston Bastiaens, CEO of Lernout & Hauspie (Belgium/Burlington, MA) who stepped up to the plate and made a commitment to the healthcare vertical marketplace. How much of this was a result of the $30 million investment in L&H by Microsoft (Redmond, WA) is uncertain, but L&H was front and center at Atlanta’s World Congress Center, making lots of interesting commitments. Why Microsoft needs L&H is something of a mystery, since Bill Gates and company enticed many well-respected voice recognition researchers away from Carnegie Mellon a few years ago and presumably has its own robust, internal voice recognition R&D under way. Microsoft is clearly about to voice-enable the Windows CE environment, although where they will find the computer CPU horsepower to accomplish this, or how they will remove the heat from such chips embedded in small, hand-held units remains a bit of a mystery. Perhaps there will be a new line of combined Windows CE computer/hand warmer devices, good for computing outside on cold fall and winter days, or during ski trips.
L&H is the Belgian-based parent that acquired Kurzweil Speech Technologies (Burlington, MA) a couple of years ago, and is now an 1,800-employee company with revenues of more than$100 million.
Christopher Force, vice president of L&H’s Medical Solutions Group, discussed new speech-enabled medical software applications, including a project to voice-enable Oceania’s (Redwood City, CA) Wave ambulatory EMR product, as well as to voice-enable medical transcription reducing the number of transcriptionists required and also reducing turnaround time. In both these areas, L&H was touting its extensive natural language knowledge bases as an enabling core technology. Such knowledge bases are not limited to the medical domain, according to L&H, but are general enough to allow foreign language translation in certain contexts. Indeed, L&H announced work to provide automatic translation in business settings from English to French to English; as well as from English to Japanese to English. Its global partnership with AsiaWorks (Singapore) resulted in the world’s first PC-based Cantonese continuous speech recognition product, dubbed SP2. It allows those who speak Cantonese to enter Chinese text by speaking continuously.
L&H has interested an array of medical companies in being its partner, including MedQuist (Marlton, NJ), Interpra (Toronto), Nine Rivers Technology (Raleigh, NC), Intelliworxx (Sarasota, FL), DVI, Lanier Healthcare (Atlanta), Computer Voice Dictation Systems (CVDS; West Covina, CA), Speech Machines and QuadraMed (Richmond, CA). It held a joint press conference with representatives from all these companies to make the announcement. All of these partners were showing or discussing voice-enabled applications at their booths.
Nine Rivers Technology will integrate Clinical Reporter into its CurrentCare Emergency Room CPR application, which handles patient registration, tracking, charting, disposition, and billing applications. Interpra Medical Imaging Networks will be integrating speech in its InterpraRadiology workflow management applications. CVDS will incorporate Voice Xpress into its DOC Solutions applications, and will jointly develop context models for pediatrics and neonatology subspecialties.
L&H predicted that voice recognition and control will become the input interface of choice for computers over the keyboard within the next five years, and implied that it will have a leading role in accomplishing this paradigm change. The company also announced enhancements to its Clinical Reporter, Clinical Reporter for Primary Care, Clinical Reporter WAV Transcriber (for voice recognition of dictated reports), and Voice Express for General Medicine.
In a related move, L&H introduced Level Assist, a software application that analyzes physician documentation against Healthcare Financing Administration billing requirements, and scores the completeness and quality of the documentation. It gives the physician a quick check of the level a clinical encounter was documented against billing documentation requirements, helping the physician retrospectively not to under-bill for the level of documentation that exists in the clinical record. This product requires Windows, running on a 266 MHz or faster PC with MMX technology, with 128 Mbytes of RAM and a 4 Gbyte or larger hard disk.
Intelliworxx will bundle voice products into its Internet browser-based mentoring applications, and announced VoiceTablet, a tablet, hand-held computer optimized for mobile voice-driven applications. It utilizes a Pentium II microprocessor in a hand-held computer with an audio subsystem optimized for voice input and output. Should be great for blind computer users.
Not only will L&H recognize your voice, but it will talk back to you in a voice you won’t recognize as computer generated. Its RealSpeak text-to-speech technology was heralded as user-adjustable, natural-sounding speech generated from text. This technology already is licensed by GTE Internetworking's BBN Technologies division.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content