Controling wheelchair by speech command using Neuron Network in labview

[duongducthien]

Hi all,

I'm doing project about "controling wheelchair  by speech command using neuron network in labview". I found some informations about controling whelchair combine with function microsoft speech recognition in window or something like that but this was not my point. So, I have some question about that:

1) How do I store the command "Left-Right-Back-Go-Stop" in labview. After that, analyzing to find the differences of characteristic of these command.

2) Based on these characteristic above, using neuron network to find exactly command.

Are there anyone do like this before? I need all help from you.

Thank you so much.

[Simon-P]

Greetings;

After doing some research on the matter, I have come across the following discussion:

http://forums.ni.com/t5/Signal-Generators/Hello-I-Want-Labview-Control-An-Indicator-By-Voice-Recogni...

In it, several users commented on speech in LabVIEW and a few VIs were posted. While I have not taken a thorough look at them quite yet, I figure they might be of aid in this application.

In what concerns the neuron network, we have an article outlining Neuron Network implementation in LabVIEW:

http://www.ni.com/white-paper/2811/en

Finally, I also found a youtube video of an implementation of a similar task:

http://www.youtube.com/watch?v=aEqa-t_TWiY

Let me know if the links help.

Cordially;

Simon P.

[duongducthien]

Thank for your reply sir.

Actually, it is just a litle bit help. However, i found out another method to control powerd wheelchair. The method follow:

Record data sound ->band-pass filter ->FFT -> save data sound as a sample (5 word: Left,Right,Back,Go,Stop are stored). After that, the user says one new word, it is automatically compare to 5 stored words by using Coefficient correlation. Which one have the value of correlation coefficient bigger than anothers, turn on the led at that position.

The result looks like good. However, I got another problem. It can not run real-time because I have to push "play back" the sound and record again when says another command.

How to record sound continuously that do not need to push "record button" again?. It means automatic cut exactly sound, there are not have unnecessary part.

Thank you for watching.

[Simon-P]

Greetings;

First, it is most excellent that you have gotten to where you can compare the words being said to the sample.

If I understand correctly, you are interested in continuously acquiring sound, saving it and then comparing it with the sample? That is not as straightforward as it sounds simply because it would be difficult to realize when there is sound in the first place and when that sound is relevant. 

I would probably start by figuring out whether there is or not sound by comparing the microphone input to a certain threshold, saving if there is sound, and then halting the recording when the sound stops. It would be then that you would compare with your samples via the algorithm that you have made.

In order to check for volume, I came across the following example:

http://zone.ni.com/devzone/cda/epd/p/id/2150

It was mentioned in the following discussion:

http://forums.ni.com/t5/LabVIEW/Microphone-Input-Detection/td-p/487367

Cordially;

Simon P.

[voicerecognition77]

Hi ,

      I am doing a project on Voice recognition wheelchair using neural networks. There are many articles that talk about obstacle detection, microcontroller, microprocessor etc. But, we are only concentrating on neural networks. Please advice me regarding this project. how do I start about it?

[voicerecognition77]

Also, this needs to be done in MATLAB or using MATLAB. Please advice me Simon

[Simon-P]

Greetings;

Generally, you might get better exposure if you create a new topic since it seems like the question is slightly different. However, for this particular matter I do not know how much help I can provide, due to the fact that this is mostly a LabVIEW Forum and I personally do not know much about MATLAB Coding. I would suggest going to a more MATLAB centric forum for a discussion regarding this application. 

Cordially;

Simon P.