Tuesday, January 29, 2013

Computers for Visually Impaired

Ideally the visually impaired would have a large screen TV or display coupled with voice recognition hardware. We are almost there but we have been almost there for many years.

Nonetheless equipment is good enough for now for many. This is the case even though voice recognition must be helped along with keyboard at times.

To test out out up-to-date speaker-independent speech recognition on an old PC, first click or tap to download the Chrome browser,  Click or tap to learn about VoiceNote, then install it from Chrome Web Market, (the last tab at the top right of the Chrome tab screen).    

The best components are out there, but not usually combined in one device. For example, the Kindle fire 7 inch tablet can drive a large screen but does not have the best speech recognition capabilities. At the same time the Nexus 7 has the latest and best speech recognition but cannot drive a large screen . 

The two capabilities need to be combined in one device. For the best results speech recognition needs to be helped along by the best technology.   That technology sends the speech pattern to a supercomputer which handles the heavy processing work, as opposed to a local computer.

It is no longer necessary to load a program into a PC, spend  hours in "training", and depend on limitations of  of very limited processing power.

The local device now simply serves as a terminal. In a small device this may require wifi. The best of this speech recognition software belongs to Google .(Click or tap for a comparison of voice search with Apple Siri.) The older Dragon system relied on the limited capabilities of an older computer and required expensive so called training of the computer to recognize only one  specific voice.

However Dragon is indeed offering better software these days For example, the iPad offers such software applications. Of course, any tablet also needs ideally a conventional keyboard to assist.

Very soon we may have more choices in devices which will work well for extensive dictation. Even so, the present systems do offer good enough performance and can be used, for example, for browser and short items such as emails. These technologies can be augmented by mature technology to read text from screen out loud.