11-14-2012 04:22 PM
Hello,
I think my problem has a fairly simple solution. However, I've searched all throughout the example codes and haven't found one that works.
My ultimate goal is to develop a home automation system that is controlled using voice commands. e.g., using voice to command the lights to turn on; to open the curtains; to lock the doors; etc. But, I don't get bogged down in the details of the items to be controlled.
My main concern is converting a voice command to a LabView output. For instance, *** if voice command = "lights on" then output = true ***
I am trying to make it as simple as possible using only LabView on a standard PC with a microphone attached.
Has anybody experimented and created a voice control VI that might work?
Looking forward to your ideas,
James Frakes
11-14-2012 04:39 PM
I haven't played with this at all, but if it helps, I personally would start by recording a command and and turning it in to an array of numbers.
Scan the incoming voice the same way - look for the amplitude to go above and below a specific threshold to make it voice activated.
Once you have a sample waveform, slide the "command" wave across it and look for a maximum match. If it matches within a certain amount (a parameter you set), it is the command. If not, scan scan the next command in the same way until all commands have been done. (By slide, I mean look at the length of the command wave, grab a subarray of that length from the command wave starting at i = 1..(max index - length of "command" array), and look for how close those two arrays match.
I have no idea if it would work, but it would be a place to start 🙂
11-14-2012 05:59 PM
Trying to do this with array matching is likely to be probablematic, and you would just be "re-inventing the wheel", so to speak. I think the best approach is to use the speech recognition utilities that are built into Windows. You will need to call the dll from LabVIEW. Here is a link to get you started:
http://msdn.microsoft.com/en-us/library/microsoft.speech.recognition.speechrecognitionengine.aspx