Millisecond Forums

Hello,

I have been experiencing some issues with the voice recognition function in Inquisit. There seems to be no clear pattern as to when it fails to record responses. Are there any general tips for improving the likelihood that the program will correctly recognize the participant's voice beyond purchasing better microphones? For example, should I set up the microphone again when running multiple sessions in a day?

Thanks.

Checking the microphone settings repeatedly surely would not hurt. However, to *really* improve recognition results, you would have to *train* the engine to each participant's voice prior to collecting data via Tools -> Speech Recognition -> Training Wizard...

But even then -- as computers are imperfect recognizers of human speech -- results will vary and recognition will (partially) fail in some cases.

Thank you very much. 

Javier Granados Samayoa

I opened Inquisit 3 and tried to run the Training Wizard, but when I did, I got the following error message:
"Speech Error 80040154: Speech Error.
Line 269, File .\SpeechEngine.cpp"

Any idea what I'm doing wrong?

If you are running a 64-bit Windows version, that's the likely the issue. Inquisit 3 is a 32-bit application (there is no 64-bit version of Inquisit 3) and hence it cannot run the 64-bit training wizard. The wizard is part of the operating system, not Inquisit. Inquisit merely "links" to it from its menu. You can run the training wizard via the Windows control panel or you can upgrade your Inquisit license to Inquisit 4 which exists as a 32-bit as well as 64-bit application.

Thank you for the help I've received so far. I have one last question. Does anyone have experience using the Training Wizard prior to conducting an experiment that uses voice recognition? Have you found that the Training Wizard made a significant difference? I've only had a small number of issues with the voice-activated Stroop task so far and I'm tight for time in the experiment, so I'm wondering whether it's worth taking participants through the training process.

Any input is appreciated.