spacer.png, 0 kB
Home arrow Blog arrow Turn your computer into a chatterbox by Paul Taylor
spacer.png, 0 kB
spacer.png, 0 kB
 
Turn your computer into a chatterbox by Paul Taylor PDF Print E-mail

In Stanley Kubrick's 1968 movie 2001: A Space Odyssey, the SS Discovery's crew communicate with the spaceship's HAL-9000 computer using natural speech.

Well, 2001 has come and gone and I still rely mostly on keyboard and mouse to work with my PC - but then, as far as I know, we have not discovered a monolith on the moon yet, either.

 

Early PC-based voice recognition systems were slow, cumbersome and too inaccurate to be much use, and many of the early commercial pioneers have gone out of business or been acquired. But in recent years a combination of software developments and more powerful computers has made talking to your machine a more realistic - though still imperfect - option.

The most powerful software can recognise tens of thousands of words, but you still need to spend time getting it accustomed to your voice and accent.

Older packages typically also required the user to speak distinctly and separate each word with a short pause. There are now several continuous-speech systems available, including the two market leaders: ScanSoft's Dragon NaturallySpeaking (www.scansoft.com) and IBM's ViaVoice, which is also now distributed and supported by ScanSoft.

The recently released Dragon NaturallySpeaking Standard 8 costs $90 while ViaVoice Standard Edition costs $50.

Both come in a variety of editions targeting different parts of the market and work with most other PC software.

For Mac users, IBM ViaVoice is the long established market leader, but MacSpeech (www.macspeech.com) offers its $179 iListen software in a UK English as well as US English version and has won a strong following despite lagging behind ViaVoice in accuracy tests. In my experience neither Mac program is as good as the PC counterparts, both of which deliver solid results.

If you were put off in the past by the serious limitations of early voice recognition programs, now may be the time to give them another look. But first there are some rules. Voice recognition software always has been, and continues to be, resource-hungry. On paper, the minimum requirements for the latest version of Dragon NaturallySpeaking include a Pentium PIII 500Mhz processor, 256Mb of memory and 500Mb of spare space on a hard drive. In practice, unless you are willing to put up with slow responses and crashes if you try to do too much at once, you really need a machine with at least a 1Ghz processor and 512Mb - preferably 1Gb - of Ram.

Skimping on specifications will only lead to frustration. It is also important to use a reasonable microphone as well as a good sound card, and to spend the time training the system properly by reading the text presented as part of the set-up procedure.

That said, I loaded Dragon NaturallySpeaking Professional 8 on to a mid-range IBM ThinkPad T42 notebook PC and was delighted with the results. Both ViaVoice and Dragon NaturallySpeaking are easy to install. But the process can take 20 minutes or more because of the need to test and (automatically) calibrate the microphone and sound card and then to train the system to recognise your voice.

I found Dragon NaturallySpeaking easier and quicker to train than ViaVoice, but once set up, both proved to be much more accurate than I had expected.

While both can be used to control a PC, switch between applications and to direct a browser to a website, most of the time they are likely to be used to create documents.

Both proved adept at this basic task - although the commands for moving text around using Dragon NaturallySpeaking are less formal than those employed by IBM. More importantly, using them with Microsoft Word, I found I could "write" significantly faster than by typing - although the dictation process is something most keyboard users will probably find tricky at first.

The more I used the software the easier it became and the fewer errors it and I made - both packages continue to learn as you use them.

Generally, I found it was easier and faster to dictate a page of text ignoring any errors and then go back and make corrections using the mouse and keyboard rather than trying to navigate the program's text-editing commands.

Interestingly, the fastest and most accurate way I found of creating new text was to use Dragon's built-in basic word processor, which is very similar to Microsoft WordPad that comes bundled with all PCs, and then cut and paste the copy into another application. I was much less successful when it came to using Dragon NaturallySpeaking with Microsoft's Excel, though this too became easier with practice.

Both packages at entry level allow you to use "natural" speech rather than the staccato dictation they used to require. The "Preferred" version of Dragon NaturallySpeaking and all versions of ViaVoice will also read text back to you, enable the import and export of files and can transcribe voice files from most digital voice recorders and from Palm Tungsten or Pocket PC handheld devices.

The top-of-the-range Professional Version of Dragon NaturallySpeaking 8 adds support for Microsoft Outlook, the ability to create macros and voice commands for other programs, and the ability to add specialised medical and legal vocabularies.

It is also designed to be installed on a network, making it ideal for small businesses with multiple users.

While most users will probably prefer to stick with mouse and keyboard, the latest versions of both Dragon NaturallySpeaking and IBM ViaVoice show that PC voice recognition software may soon provide a viable alternative interface for those with special needs, or who feel more comfortable "talking" to a PC than navigating a keyboard.

This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

 
spacer.png, 0 kB
spacer.png, 0 kB
2011 - Centaur Academic Media - design by Joshua Arciniega spacer.png, 0 kB