-4

I am doing a project on developing a silent speech interface using eeg sensors. The equipment available is ADI powerlab kit which allows the use of three to four elctrodes maximum. My question is can i sense the speech using this equipment or do i need to acquire an electrode cap.Also what are the nminimum number of elctrodes which are needed for this project?

Scott Seidman
  • 29,274
  • 4
  • 44
  • 109
newbie
  • 19
  • 1
    This question is better addressed to electrophysiologists or neurologists than electrical engineers. – Matt Young Oct 23 '12 at 19:12
  • 1
    Hit the literature. See what has already been done for EEG speech detection, and how it has been done. – Nick Alexeev Oct 23 '12 at 19:36
  • Perhaps "How can you process the signal from N EEG electrodes to identify speech, and how big does N need to be?" might be more on topic. – Scott Seidman Oct 23 '12 at 22:31

1 Answers1

2

My specialty is bioinstrumentation, and I teach physiology (but not bioinstrumentation, where I use bench test gear) with the ADI systems.

My bet is that you will not be able to do this, but it might seem like you can! EEG is small, noisy, and has a good deal of randomness. The way you deal with this is to take an ensemble of trials time locked to a specific stimulus and average them. Noise and any randomness get attenuated by the square root of the number of samples. This is the method behind auditory evoked potentials (http://en.wikipedia.org/wiki/Brainstem_auditory_evoked_potentials ). I doubt very much that anything you'd be able to get anything useful without really advanced brain-machine-interface processing techniques.

Why do I say you'll think it's working?? I think you'll train yourself to make very subtle scalp movements (without even being aware of it!) and EMG will mix with EEG to do whatever you want to do. I have no clue how you can control for this.

All that said, check out http://theneuronetwork.com/group/braincomputerinterfaces, and look at the reference list. Wolpaw is highly recommended, and has been key in this field. Many people with high spinal cord injuries have been following this sort of a system for decades hoping for the next big step.

There's also http://neurogadget.com/2012/06/18/openvibe2-offers-open-source-brain-computer-interface-platform-ubisoft-supports-the-initiative/4485, though I really wonder about EMG contamination, and would really look at the guts of how they're processing EEG before I'd put a seal of approval on something like that.

Scott Seidman
  • 29,274
  • 4
  • 44
  • 109