05 Mar Tech upstart VoVision could disrupt the voice-recognition market
Madison, Wis. – Paul Hager admits the technology associated with his digital product isn’t perfect, but he’s found proprietary ways to address its flaws and transform it for the voice-recognition market.
Hager, co-owner of an early-stage Madison business called VoVison, has ambitious plans for his Web-based, voice transcription technology, which has three differentiating features: a hosted solution, plus self-correction and the ability to work with multiple voices.
The technology’s feature set, which is part of a patent application, promises to accomplish what previous voice-recognition technology could not – enable businesses to save, search, and organize voice mail, and reduce the transcription process to a simple spot check.
It is something that Hager, and many other business consumers, once assumed was coming down the pike.
His hunch, which was born of a technological and legal background, was based on the fact that computer processors were getting faster and faster, and that executive secretaries were still spending a lot of time transcribing dictations and voice mails.
Hager, who took occasional computer science classes while pursuing a legal-studies degree at the University of Wisconsin-Madison, found that his assumption was premature, and that convinced him to work on a more advanced voice-recognition product.
“That voice recognition, while it’s a very good idea and every year they say it’s more accurate, it never actually became accurate enough,” he explained. “That’s where people got this frustration with voice recognition that has caused people to not really embrace it, even though it’s embedded in the Windows operating system and has been for a number of years.”
Hager felt that people would have seen the business value if those promoting voice recognition shifted their focus from falsely claiming that it’s 100 percent accurate, and instead concentrated on developing a self-correcting and voice-independent system that could be a reliable business intelligence tool.
Hager believes that day has arrived with a voice-recognition product that will be marketed to the legal, healthcare, insurance, and government sectors. It has no official customers yet, but the product is being tested in the Osceola County (Florida) District Attorney’s office, and may soon be adopted by the University of Wisconsin School of Medicine and Public Health.
Playing host
To access the hosted software, all a user needs is a computer, an Internet connection, and a phone – specifically, a digital-based phone system such as Voice over Internet Protocol, a hosted PBX, or a device where voice-mail files can be stored in a digital file format, such as a wave file or an MP3 file.
The system uses that data to perform the transcription, and there is a Web interface that users log in to listen, correct, and interact with the system, which includes transcription servers. The infrastructure’s back end contains a program where the system trains itself to embed any corrections for greater accuracy in subsequent uses.
The most obvious benefit is in time-savings. An experienced court reporter, even one that can type 80 or 100 words per minute, will take about two hours to transcribe an hour-long recording. According to Hager, VoVison’s computer can read that one-hour audio file in 20 minutes, and the rest of the work involves a short, spot-checking process to determine how much correction is required.
Hager’s business plan is not to sell the product independently but partner with technology vendors that would offer it as a value-added service to their customers, and pay VoVision a monthly subscription fee. After growing each market to at least $10 million and perhaps to as much as $100 million annually, his eventual goal is to be acquired by a large carrier such as TDS or AT&T.
Voice to Vision
VoVision – aka from Voice to Vision – is a partnership of Hager and Madison attorney Linda Balisle, a professional relationship that began when Hager clerked at the Balisle & Roberson law firm while doubling as a college student. The firm eventually hired Hager as its IT manager and, ironically, Hager created the technology in part because the firm did not use voice mail.
Balisle said neither partner is afraid to ask hard or “seemingly stupid” questions, and in a statement that admittedly dates her, Balisle said she admires Hager’s lack of fear to “go where no one has gone before.”
She did, however, have a pointed query amid Hager’s initial excitement over the new technology: “So what does that do for me?” she asked.
While Balisle & Roberson has yet to completely integrate the technology, it already has paid dividends in faster dictation and corrections. Balisle said it has eliminated the need for dictation equipment and the need for a professional transcriptionist, plus it’s faster and it trains itself to work with multiple voices.
The more people use it, the more they benefit from the corrections of others, as VoVision proved in a live demonstration of medical terms. Balisle was interested in seeing what the system would do when users entered a medical term that no voice transcription or recognition system would have heard before – in any voice. The true test was to have a previously unfamiliar word go through, have the product correct it once, and see what happens when the term is used again in different documents.
Balisle was delighted to see that when new terms were used in medical dictation, it only took one correction for other dictation documents to use the same word – a feature that could make the transition to electronic medical records less problematic.
“The big problem that we’ve been blighted with is when you have your voice-recognition system, it only worked with one voice,” Balisle noted. “Ours works with all voices. Once our transcription system and correction system has done a particular voice, a particular sound, and sees what it’s supposed to say, then everybody that uses it gets the benefit of that correction because all the corrections are made through the same database.”
The digital-analog connection
At the time they decided to partner, Hager was only 21 while Balisle has been an attorney in Madison for 27 years. Her friends and family often ask her why she would partner with such a young entrepreneur.
“My answer was and is, `His idea solved a problem, was elegant and brilliant, and as we approached the engineering and design step by step, we were able to address and solve every obstacle along the way,” she said. “I think a young digital brain like Paul’s, combined with a mature analog brain like mine, can create remarkable results.”
Related stories
• Paul Gibler: Lights, cameras, action: The state of online video
• Johnson Controls partners on car satellite linkup
• Digitizing hospitals with the right tools
• Merge eFilm announces rapid growth in first quarter
• API Releases New Software, Acquires Additional Contracts