Voice recognition enters limelight as one-time powerhouse crumbles

Evolving applications focus on workflow and integration

When Lernout & Hauspie acquired Dictaphone and Dragon Systems in early 2000, the sky was the limit for the Belgian firm. L&H was going to unify the healthcare industry with the implementation of enterprise-wide speech recognition and transcription. Now, two years and many revelations later, L&H is being carved into pieces.

The company, which declared bankruptcy about a year ago, lost its protection under bankruptcy law at the end of last month. Since then, the pieces of this once-respected firm have been flying. Its translation business has been sold to Bowne & Co. for $44.5 million and its medical transcription business to Medquist for $24.8 million. Dictaphone will become an independent company again, most likely by year-end. U.S. and Belgian courts will oversee the final auction on Nov. 26 of the remaining assets, including the core speech-recognition technology.

Ironically, the promise of voice technology has never been stronger.

“In years past, customer adoption in the voice market was slow, as in PACS,” said Michael Mardini, president of Talk Technology in Bensalem, PA. “This year and next will be coming-out years for speech recognition. Customers calling us now aren’t looking just for a demo; they want to talk about technical aspects and implementation.”

Talk Technology builds workflow applications that can be embedded in any core speech engine. The company has also integrated its Talk Station product into several PACS. And as an added bonus, the company will feature products at the RSNA and Healthcare Information Management Systems Society meetings that show how radiologists can go mobile with speech recognition, Mardini said.

“Dictate, edit, and sign anywhere-that’s where companies like ours are going,” he said. “At RSNA we will show how radiologists will be able to use handhelds to open, edit, and sign reports. We’re making our entire suite of applications more useful to clinicians, and providing better, more efficient ways to generate reports and get easy access to information.”

IBM’s Voice Dictation Products division is also experimenting with handhelds. Big Blue is working with handheld companies to integrate speech recognition and dictation into personal digital assistants, said Toby Maners, segment manager for Voice Dictation Products. The company plans to launch its first voice-based “command and control” product on the Compaq iPAQ in the next few months.

“We’re seeing doctors who want to use their handheld devices for dictation,” Maners said. “Because of increased processor power, availability of communications bandwidth, and wireless networking, physicians can dictate into their iPAQs, send the audio file to the dictation server where it will be processed by the speech recognition engine, and then receive the text file via wireless e-mail.”

Despite advances in hardware and software technologies, Maners doesn’t believe the transcriptionist will ever be completely replaced.

“A person must always verify that the text that comes back is accurate compared to what was dictated, and we don’t want to dump the editing on the doctor,” she said.

John Donohoe, president and COO of Medquist, agrees. Marlton, NJ-based Medquist is channeling its development efforts into implementing speech recognition software in its own workflow to reduce labor costs.

“I think that voice technologies will ride our back into the marketplace,” Donohoe said. “Our strategy is not so much to sell speech recognition products commercially, but to use them behind the scenes.”

The company expects to roll out a product in the third or fourth quarter of 2002 that will leverage voice recognition technologies on the back end. The result will be that transcriptionists will function more as editing and quality-assurance staff, thereby increasing productivity and potential cost savings for customers, as well as Medquist’s margin.

Medquist will also continue to market its doc-in-the-box Many Speech software. This stand-alone package for subspecialties such as radiology and cardiology allows the user to either self-edit reports or route files to the Medquist network so that Medquist staff can correct and edit the report.

Medquist completed the purchase of L&H’s Medical Solutions Holding on Nov. 13 for $24.8 million. The firm will integrate the U.S. divisions of the transcription services company, according to Donohoe, but will either close down or sell off the Philippines transcription services. More acquisitions and expansion into other horizontal and vertical markets, such as legal, financial, and data mining, are part of Medquist’s long-term strategy.

Former L&H division Dictaphone is also aiming at a larger market. The firm should exit from Chapter 11 and regain its former status as an independent company by the end of the year, which puts it in an ideal position for growth, according to Don Fallati, senior vice president of marketing and strategic planning for Dictaphone. Its Healthcare Solutions Group accounts for about half of the $250 million company.

“We have lost very few customers during the last year, though some did delay purchasing while waiting for our situation to be resolved,” he said. “We are pursuing growth. The labor cost alone in transcription is $6 billion to $10 billion per year, and speech recognition takes direct aim at that cost and at bringing productivity to healthcare.”

Dictaphone, based in Stratford, CT, acquired PowerScribe for Radiology from L&H in October and also licensed MREC, one of L&H’s core speech recognition engines. In addition to PowerScribe, the company will showcase its partnerships and integration with RIS and PACS vendors at this year’s RSNA meeting. PACS is another channel to drive speech recognition into the radiology department, according to Fallati.

But the path does not end in the radiology department. Dictaphone’s Enterprise Express (EXSpeech) software builds on the dictation transcription management system and brings speech recognition to the rest of the hospital using telephone-based input.

“We believe that the Dictaphone brand name, history, and understanding of physicians and workflow put us in a prime position to attack the broader marketplace,” Fallati said. “If we really want to take the benefits of speech recognition to a wider audience, we have to move out of the departments.”