Speech Recognition Software

[Labviewguru]

I am considering building applications that use speech recognition as an input form. The only things I can find out there that look decent are IBM's ViaVoice and Dragon's NaturallySpeaking, which I have heard some good about.

Two things: one, which is, by everyone's experience, a more reliable and flexible recognition software (IBM? Dragon? Other??). Second, who has the best development system. I have seen others that use ActiveX for communication to other software, which is likely useful in LabVIEW, but I would prefer to access dlls for this.

There are two purposes I have, actually. One is to find a good speech recognition package that I can use for development, and for extensive writing. The other is to develop applications in
LabVIEW that accept speech input.

Please respond with your opinions and experiences.

I would be very grateful.

Thank you.

[Enrique]

Labviewguru;

Both IBM's ViaVoice and Dragon's NaturallySpeaking are very mature speech recognition softwares. You may want to look closely at NaturallySpeaking because the company is currently experiencing some changes.

An alternative is Microsoft's Speech API (SAPI), which is free (compare with ViaVoice SDK - aprox. $1k - although is not for nothing they are leaders in this area.) The current version is 5.1. It is only for the english language and it recognize lots of the things I say, even when I have my "espa?ol" accent. The more you train it, the better it perform (duh!).

I know somebody made some VI's with SAPI (version 2 ??), but only for text-to
-speech. I was unable to make them work, and I don't know of they are still available.

I have SAPI v 5.0 and and created a simple dll which I call from a VI to do text-to-speech. Now my computer can recognize me and call me by my name with LabVIEW 🙂

However, in trying to do speech recognition, I found the main problem is to make your VI to "listen". Multithreading is key here. It's been a while since I did my little experiments with speech recognition, though.

I'll let you know as I get more info.

Enrique

(NOTE: I am NOT related to Microsoft nor endorse them. Just in case.)

[Lab Viewer]

Hi,

As far as I remember, in Europe, PHILIPS is quite appreciated for their voce recognition package. I know that they have a great package, especially for developers/integrators (with both C API and ActiveX access). A full working evaluation for 90 days is another good point. [http://www.speech.philips.com/pc/pages/pc_home.htm]
I a friend told me that is quite good, but unfortunately I didn't have time to try it.

Good Luck!

[Rebecca]

Hi,

I have very recently worked with LabVIEW and speech recognition. I am using the Base 6i package. Even so, Microsoft Agent was able to "understand" and listen for the words I programmed into an array. In my playing with this feature, I programmed questions that the computer (after recognizing through listening)would answer (through text to voice.) For a free download, I think Microsoft agent is a very good place to start and familiarize yourself with the process. If you would like elaboration, let me know. Hope this helps!

[markwysong]

Rebecca,

Could you email me about this application? I'd be very interested in it. Thanks.

Mark

mark.wysong@ae.ge.com

[Enrique]

I just discovered the following in Developer's Zone:

Voice Recognition in LabVIEW

Regards

[Labviewguru]

Follow up on my experience with Speech Recognition:

I am currently evaluating a borrowed copy of Dragon 4. I find it very useful, but a few flaws I have noticed:

1: Its slower than molasses in January. I have not found this to be a productivity increaser, and have abandoned it for that effort (I am writing a LabVIEW book, and this doesn't make it easier.)

2: Other than being slow, it works extrememly well.

3: The 'training' is incorrect, in my opinion, in implementation. You are training it by reading something. I have a different voice, tonality, and expression when I am reading, versus when I am dictating from memory, or just talking. This is how the training should be done. I am going to retrain it by memorizing the passages it shows for
training. This makes it more difficult for me, but should increase the productivity. Going forward, I hope developers see this, and are able to compensate in their training.

4: I see a lot of application of SR in LabVIEW. However; it will be a long time coming, unfortunately. I am going to try to work on this. If anyone else is or plans to work on SR, please initiate correspondence with me.


If anyone has any other experiences with SR, please let me know. I think this is the future (Gene Rodenberry was right on the mark once again.) I have plans for an AUI (Audio User Interface.) I think LabVIEW would be a good testing ground for this. My plan is to use LabVIEW for Linux, as there is already too much overhead in Windows for it to be effective.

Any ideas?

Thanks,

Mike Du'Lyea

[Tom Qi]

Hi there,
Is there a small vocabulary speech recognition programmed in LabVIEW?
I look forward to hearing from you.
Regards, Tom

[Brian_K.]

Hi Tom,

I did some searching around and could not find any out-of-the-box example for speech recognition in LabVIEW.  It looks like most people who try speech recognition use a 3rd party DLL or something along that line.  What are you trying to do, and we can see if anyone in the community has done it.

[cpalka]

Dear all

I have a board connected to a encoder to measure the distance and it save into a file.

I want to add into this file some voice commands thar are recognised and saved in the same file as text.

I explain, i go with a car measuring and saving the distance but if i find some signal i speeak into a microphone like "stopt" it recognises and save the word in the same file and at correponding distance.

Best regards..CPalha