When lawyer kelly Webster in London, Ont., does her correspondence, she forgoes her keyboard, puts on her headset and talks to her computer.
Thanks to speech-recognition software, her words appear as type on her computer screen. Using voice commands, she can create documents, edit and print them, surf the Internet and send e-mail — all without touching a keyboard.
More important, Kelly says, “We believe we can give our clients better service and can move more quickly and responsibly, and be more cost-effective.”
Financial advisors wanting to escape the drudgery of producing client correspondence can take a page out of the legal community’s book and consider adopting speech-recognition technology.
The technology has caught on in law and medicine; it is used by online brokerages and banks to handle telephone inquiries, but it is still finding its way with financial advisors, say the experts who make the technology.
Brian Garr, program director of contact centre solutions at IBM Corp. in White Plains, N.Y., which makes the voice-recognition program ViaVoice, attributes this to the fact that, in many brokerage houses, people work in cubicles in a large, open area. “People may feel strange about speaking out loud,” he says.
Matt Revis, senior product manager of dictation products at ScanSoft Inc. of Peabody, Mass., which makes the popular Dragon NaturallySpeaking voice-recognition software and is the global distributor of ViaVoice, says financial advisors stand to gain from using this technology. That’s because they produce a lot of documents and work with a range of software programs, most of which can be adapted to be used with voice-recognition programs. “Anywhere you see a blinking cursor, you can dictate,” he says.
Dictation, Revis notes, can be much faster than typing: “Most people talk three to four times faster than they type.” ScanSoft research shows that the average typing speed is 40 words a minute, meaning it takes about 22 minutes to produce a 900-word document. That same document can be dictated in six minutes, as people speak English at about 140 to 160 words a minute.
Voice-recognition software has been around for almost a decade. The early versions, however, left something to be desired. They required users to speak into their headset microphones in a staccato fashion. As well, accuracy was an issue in the early versions.
Distinguishing between statements such as “I scream” and “ice scream,” or “principal” and “principle” was a challenge. So was coping with background noise, which the microphones often picked up, making some words illegible on screen.
Now, however, voice-recognition software is 99% accurate. “There has been a substantial improvement in accuracy,” Revis says, adding the software is becoming extremely popular. Voice-recognition software works off phonemes, the smallest units of significant sound in a language.
“Anytime you articulate a phoneme, it makes a unique shape,” he says. The software interprets that and strings the sounds together into syllables and words based on probabilities.
ViaVoice and Dragon are the two leading voice-recognition systems. They are easy to use and load like other software programs.
They come with state-of-the-art microphones and headsets designed to eliminate background noise and focus on the sounds made by the person speaking rather than background noise. But they are not foolproof, Revis says. If there is a jackhammer operating in the background, he says, it will affect performance.
Both ViaVoice and Dragon come in a variety of packages, so you should select a package that suits your needs. Prices are also all over the map, ranging from a low of $32 for ViaVoice’s bare-bones consumer package to a high of $980 for Dragon’s professional version, which is probably more suitable for advisors.
But be forewarned. The programs can be memory hogs and require a fair bit of computer power. Dragon likes 500 megabytes of free disk space, while ViaVoice requires 510 megabytes. Once the program is loaded, users can choose between dictation mode or a blend of dictation and the mouse.
For example, instead of navigating your computer or the Internet using a mouse, spoken commands can be used to open programs such as Microsoft Word and Excel, to create text in them and even, in the case of Excel, to run calculations. You can also use the programs to navigate or create e-mail.
Here are the different versions, starting with ScanSoft:
Turning the spoken word into text
Speech-recognition software can save advisors time, money — and serve clients better, proponents say
- By: Jim Middlemiss
- May 31, 2005 May 31, 2005
- 09:15