Voice Recognition with LabVIEW and Windows Vista

[HoSsEiN]

Hello All,

I have been searching about the ways to do voice and speech recognition with LabVIEW and came across this link:

http://www.ni.com/swf/flv/labview/us/vista/vr/

I do use Vista and I have LabVIEW 8.2 (Vista Capable).  I tried to find the vi that he is using here (Speech Recognizer - Fan Demo.vi), but no luck so far.  Does any body know how I can find an example like this?

Your help is greatly appreciated!
Regards,
HoSsEiN

[jbrohan]

Hi HoSsEiN
I'm in a similar situation and maybe we can combine resources to solve this problem.
I 've downloaded the Microsoft 5.1 voice SDK, it's apparently free. This contains some C++ examples which in one case I've successfully compiled and got to work.
I had no luck with the NI 6.x examples from the NI Code Sharing web location.

My interest is to interpret a small (10-12 word ) vocabulary from cell phone recordings. I have all the technology for extracting .wav files from cell phone voice mails.

The SDK claims to be able to use .wav files as input.

My particular difficulties are to interpret the C++ example and convert it into LV and then to get the file input working.

Did you manage to get the NI Voice Recognition software working? http://zone.ni.com/devzone/cda/epd/p/id/4401

I'm using Windows XP and LV 7.0

John

[Daniel_E]

Hey,

The example that is being used in the webcast uses .net constructor nodes, property nodes, and invoke nodes in LabVIEW to call an assembly installed with Vista.  The assembly is System.Speech, I believe.  I also believe that this assembly exists in XP as well.  So, the functions being used in the demo are already predefined by Microsoft and just being called through LabVIEW.  Let me know if you need any help setting up the .net stuff in LabVIEW.

Hope this helps!

[jbrohan]

Hi Daniel

This would be really (really, really) useful. If I could make a suggestion...A very simple speech recognition, with just a loop listening and displaying. This would give me a startiing point. .net is a bit of a mystery to me.

If you have the knowledge I'd appreciate ointers to how to specify a particular dictionary, and how to specify file input rather than microphone.

Since I'm stll on LV7.0 could you post a jpeg as well s he code?

This simple example can be readily expanded to get menu items from teh speech as well as a menu, which would be kind of fun!

jbrohan@gmail.com

[FMonkey]

I've written a small example that might help get you started.  The first sets up the phrase choices, and the second is the callback that adds the result into a string control.  The last step could be to have an event structure trigger when the contents of the string control (referenced in the callback) changes.

[HoSsEiN]

FMonkey,

Thanks very much for this example.  I had played with this .NET API last night, but I couldn't have the right setup for it.  I am sure this will help a lot, so thanks again.

Regards,

HoSsEiN

[FMonkey]

Hers is an excellent MSDN article that might also help.  Look at the code (figs 8&9) under section System.Speech.Recognition.  It explains how to create more complex grammers. Link

[jbrohan]

Hello fMonkey
This is great.
Could you publish the program too. I'm sure I can get to an LV 8.x and try and get it to work. Does it work only on Vista?
The sequence is clear, even if I'm not at all clear how it starts off with "Choices" on the left. Maybe I should update my LV!
Thanks John

[FMonkey]

Attached is the initiation subvi.  I've moved things around a bit in hopes to keep the logic clear.  It might be easier for you to think that "SpeechRecognizer" goes first.  "Choices" is just the first step in building a "grammar".

As for for your question if it requires Vista, the short answer is yes. While you can install the .NET 3.0 Framework on XP, some features are specific to SAPI 5.3 which is only available on Vista.

[jbrohan]

Can I share my attempts to get Voice Recognition going in LV7.0 and Win XP.

From studying the example ms Agent User Interface Example.vi  in the Code Sharing part of NI Developer Zone

The following are the conclusions I've reached.

1. You need to create a character to do the listening or speaking.
2. You can register an Event Callback on "Listen Complete" to the Axtivex Object MS Agent. It creates an empty callback routine if you click on VI Ref in the Reg Event Callback node
3. The Callback VI should catch and show the words it recognizes.

I do not know how to control its dictionary.
I do not know anything about the character LV6i.acs (default from example) Is this where the dictionary is located?

I have to stop now. I'll look with hope and interest to any contributors to this thread.

John