Windows speech recognition frequently recognizes the commands home and move when I am not speaking.

This problem makes it difficult to surf the Internet because I am frequently returned to my homepage.  It also interferes with dictation because the cursor is moved around at natural pauses and at the end of sentences.

I’ve tried preventing these words from being dictated in the speech dictionary, but that doesn’t seem to stop them from being interpreted as commands.  I don’t know a way to correct windows speech recognition when it makes this mistake, because they’re commands.  Can anyone make any suggestions?

Thank you very much for all of your suggestions.  I took some time to try them out and see if they helped over the long term.

I have begun running the utility, " set up my microphone" several times a day, as the volume and activity in my office changes.  That has had the greatest effect at reducing this issue.

I have the microphone off the corner of the mouth, and have been experimenting with how far to the side I can place it before that "setup my microphone" utility tells me that it cannot hear me clearly.  This has also had some positive results.

I don't hear any electronic noise on the line, and did not have a different experience using other USB ports.

I took the basic step of removing the Home button from my browser, and use notepad to dictate any longer web site posts, such as this one, to avoid accidentally navigating away.

I still have some issues with these commands being issued, but only at the rate of one every couple minutes when I am sedentary.  When I am actively moving around I have more issues, so I try to avoid doing that at the same time as voice recognition.  I have even better results when I drop the pitch of my voice about an octave, but it is not realistic for me to speak that way for long periods of time.

I am curious what level of accuracy I should expect here, in terms of words and commands not being received when I am not speaking.  I am very happy with the accuracy when I am actually dictating, and think that it is above 90%.  However, I haven't read anything that tells me what to expect in terms of filtering out sounds that are not speech.

Thanks again for all of your help.

Thanks for the response.

I can confirm that despite what I set the audio settings to, they jump up to 100 after running the 'setup my microphone' utility.  I am using Windows Vista.

I personally have not seen any negative effects from this change- my accuracy seems to be about the same as it was before.

However, my work space has also been much quieter lately, as this is a time of year when many coworkers are on vacation.

Sarah

Sarah wrote: I can confirm that despite what I set the audio settings to, they jump up to 100 after running the 'setup my microphone' utility.  I am using Windows Vista. I personally have not seen any negative effects from this change- my accuracy seems to be about the same as it was before.

I modified the SNR key (hit the decimal radio button) to 32. The Volume key (decimal again) was set to 3333 (33%). These seem to work well with my home office setup (a Samson Q7 and Andrea USB pod), my Toshiba Satellite notebook (Sennheiser ME3/Andrea USB pod) and my business office (Sennheiser MD431II/Andrea USB pod). Remember, once you reset the SNR and volume, do not run the setup my microphone audio setup wizard. If you do, it will set the volume back to 100%. Also, do not forget to reboot your computer for the settings to take hold. Finally, of course you are going to make a registry backup before playing around?

SNR setting:

HKEY_CURRENT_USER\Software\Microsoft\Speech\RecoProfiles\Tokens\{95CF724E-B3B5-4D94-A4FB-36AE77A88FE0}\{DAC9F469-0C67-4643-9258-87EC128C5941}\SNR

Volume settng:

HKEY_CURRENT_USER\Software\Microsoft\Speech\RecoProfiles\Tokens\{95CF724E-B3B5-4D94-A4FB-36AE77A88FE0}\{DAC9F469-0C67-4643-9258-87EC128C5941}\Volume

Marty Markoe, eMicrophones, Inc.