Jan 11 2009

Windows 7 (b7000): Speech Recognition Testing

I’m not used reporting this, but I’m pleasantly surprised with the changes that Microsoft has made to the speech recognition portion of the operating system. This information is based upon the windows 7 beta version 64 bit.

Like Vista, there’s not really an installation For speech recognition in Windows 7; It’s more like select the option choose some settings and start using the application. Locating speech recognition in Windows 7 is very easy, The first thing that you need to do is go to the start menu in the bottom left hand corner of the desktop, then you simply type into the run box ‘speech’, the results will be displayed for you in the menu and you simply choose ‘speech recognition’.

Read more after the jump…

To begin using Windows 7, Like Dragon Naturally speaking, there’s no training involved [but you can if you want] to be able to start using the application immediately. The operator only needs to choose the option if they’re using a headset, if they want to use document review [automatically adding words to a vocabulary] or if they would like to activate the advanced features of the speech recognition to control the operating system almost completely by voice.

Windows 7 seven has improved the user interface for speech recognition. It’s not perfect by any means, but it does allow for the operators to interface with applications that are not specifically written four speech recognition with Microsoft Windows. For example, if you dictating into a Microsoft product or application the speech recognition seems to work flawlessly, but if you’re dictating into application that’s not updated or is not the standard with what Microsoft considers compatible, although spoken text is placed into a text box and then when you’re finished speaking you simply insert the information.

sbres-1231702113-0 Windows 7 (b7000): Speech Recognition Testing

This is a huge improvement for people who are trying to free themselves from costly Dragon NaturallySpeaking or transitioning from Vistas version of speech recognition. The way the new interface works, the operator will speak in a normal tone and the application will pick up on what the user is saying, the recognized words are placed into a small text box. Information is not immediately pasted into the destination application, it’s held in a small box for review and then any changes or corrections are made inside the box and then the operator simply has to say Insert.

An exception to this [and very disappointing] would be the free office package, OpenOffice. Version three of OpenOffice is not compatible with Microsoft’s voice recognition software. Of course this is a beta version and things could change by the time the final release of windows 7.

The performance for Windows 7 is showing to have a good response time. In the course of my testing the speech recognition of the applications seems to be in the 90 percentile range. The speech engine and resources seem to use almost half of what dragon naturally speaking uses. Dragon naturally speaking typically uses anywhere from 113-170 MB of memory while in use. There seems to be an obvious improvement here with System Resources as it pertains to speech recognition.

The system seems to be able to seamlessly pick up between users. One person can be using the microphone at one point, speaking sentences and then the other person can switch to using the same headphones and be able to be recognized properly by the speech recognition software [I tested this]. I believe this to be another improvement, but they’ve made to where speech recognition software could be would be highly productive as it pertains to fields where individuals are professionals who are working together and need to be able to use the same machine, quickly and enter lots of information in to a field and not have to switch user accounts.

I have reported before that if Microsoft gets their stuff together, it’s really going to put the squeeze on Nuance to keep charging those outrageous prices for their software. This was a nice step forward. I look forward to a final release.

My suggestion for Microsoft at this point is to try to open up more.  Be more compatible with software packages, don’t be such a closed shop when it comes to speech recognition.  Speech recognition should have the ability to have the assignment for hot keys; turning on and off with the option for corrections.  And before the final release is made my suggestion is that Microsoft should make the adjustment that when the text box is full that if no corrections have been made inside the text box the information should be simply pasted into the application allowing the user to continue as opposed to being interrupted with the options to insert; whether by clicking a button or saying the word insert.

This document was created entirely with Windows 7 and it’s Speech Recognition Engine.


Support the site — Share this!