Print

Mail

wiki links

Speech recognition makes keyboards a thing of the past

Microsoft's new Windows Vista operating system comes with integrated speech recognition software. This allows users to perform standard tasks without using their keyboard, including writing e-mails.

Those who find the hands-free life enjoyable may want to consider more comprehensive specialist software, however.

Kolokythas Panagiotis has already spent some time talking to his computer.

"When I'm not in the mood to click on the Start button with the mouse and look for the program, I just say `Open Word' or `Write Mail,'" explained the editor from Munich-based PC Welt magazine.

Users must first complete a one-hour acclimation program before the computer can understand commands. Vista also runs a piece of software in the background to handle speech recognition tasks.

Vista's speech recognition works relatively reliably, the experts claim. It is not really suited to professional users, however, meaning that users who frequently dictate texts as part of the working day are better served with special applications.

For one, higher-end software requires no break-in time, as the "software secretaries" come ready to handle various voices and dialects. Indeed, when creating "Voice Pro," now in its 11th version, Linguatec invited numerous people into the IBM development lab to read texts aloud.

"We made sure that the speakers came from various regions where the language is spoken," said Rieke Bacher, a computer linguist employed by Linguatec.

Each word is compared with a pattern known to the system, explained Martin Held, a developer at Nuance in Munich.

"If the system is confronted with a word it doesn't know, then there can be no recognition," Held said.

That means that no recognition software will ever be 100 percent effective.

It's an entirely different matter in situations where there is only a limited pool of vocabulary.

"Doctors or lawyers do very well when dictating within a restricted set of topics," Held said.

Those target groups are even offered programs pre-loaded with vocabulary pertinent to their discipline.

Yet even in cases where a computer is faced with highly unusual jargon, such as an archeologist, it's not a lost cause.

"The vocabulary analysis tools can take your written texts and analyse them to establish your personal vocabulary," Bacher said.

That applies to "Dragon Naturally Speaking 9" from Nuance as well.

Both "Dragon Naturally Speaking" and "Voice Pro" work with digital dictation devices and wireless headsets. This lets drivers talk while on the road, with their spoken notes automatically converted into writing.
This story has been viewed 1528 times.